# Source paper archive

- Source title: Genomes to Fields 2026: Genomic resources for investigating genotype-phenotype relationships in Zea mays
- Author(s): Johnson, N. A., Murithi, A., Grover, C. E., Kaeppler, S., Lima, D. C., Bhayani, H., Lokuwalpola, D., Chen, H., Allsing, N., Michael, T. P., Buckler, E. S., Hufford, M. B., Romay, M. C.
- Source landing URL: https://www.biorxiv.org/content/10.64898/2026.11.15.766802v1
- PDF URL: https://www.biorxiv.org/content/10.64898/2026.11.15.766802v1.full.pdf
- PDF SHA-256: c1e23ec82b0f0edf51b4fee9660ab0f48a9889b09d2da1db27b57c7bb018875b
- Converter: pymupdf4llm 0.0.27
- Limits: Automated PDF text extraction; reading order, equations, tables and formatting may be incomplete or incorrect. Images are omitted, and no OCR was performed. Markup and links are neutralized for safe downloading. Consult the original PDF for the authoritative version.

---

&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.15.766802; this version posted November 18, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.15.766802)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


1 **Title: Genomes to Fields 2026: Genomic resources for investigating genotype–phenotype**
2 **relationships in** _**Zea mays**_
3 Authors: *Nicholas A. Johnson &#91;1&#93;, Ann Murithi &#91;2&#93;, Corrinne E. Grover &#91;2&#93;, Shawn Kaeppler &#91;3,4&#93;, Dayane
4 Cristina Lima &#91;3&#93;, Hirva Bhayani &#91;2&#93;, Dinuri Lokuwalpola &#91;2&#93;, Heidi Chen &#91;5&#93;, Nicholas Allsing &#91;5&#93;, Todd P.
5 Michael &#91;5&#93;, Edward S. Buckler &#91;1,6&#93;, *Matthew B. Hufford &#91;2&#93;, *Maria C. Romay &#91;1&#93;
6 1. Institute for Genomic Diversity, Cornell University, 130 Biotechnology Building, NY 14853
7 USA
8 2. Department of Ecology, Evolution, and Organismal Biology, Iowa State University, Ames, IA
9 50011 USA
10 3. Wisconsin Crop Innovation Center, University of Wisconsin – Madison, 8520 University
11 Green, Middleton, WI 53562 USA
12 4. Department of Plant and Agroecosystem Sciences, University of Wisconsin – Madison, 1575
13 Linden Drive, Madison, WI 53706 USA
14 5. The Plant Molecular and Cellular Laboratory, Salk Institute for Biological Studies, La Jolla,
15 CA, 92037, USA
16 6. USDA-ARS Plant, Soil and Nutrition Research Unit, RW Holley Center, 538 Tower Rd,
17 Ithaca, 14853 NY USA
18 *Author for Correspondence
19

20 **ORCIDs:**

21 Nicholas A. Johnson: 0000-0001-7272-6474

22 Ann Murithi: 0009-0007-2229-2656

23 Corrinne E. Grover: 0000-0003-3878-5459
24 Shawn Kaeppler: 0000-0002-5964-1668
25 Dayane Cristina Lima: 0000-0002-5327-2160
26 Hirva Bhayani: 0009-0005-4483-030X
27 Heidi Chen: 0000-0003-1744-8461
28 Nicholas Allsing: 0000-0002-1145-1646
29 Todd P. Michael: 0000-0001-6272-2875

30 Edward S. Buckler: 0000-0002-3100-371X

31 Matthew Hufford: 0000-0003-3945-1143
32 Maria C. Romay: 0000-0001-9309-1586
33


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.15.766802; this version posted November 18, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.15.766802)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


34 **Abstract**
35 Maize ( _Zea mays_ L.) is a major global crop characterized by high genetic and phenotypic
36 diversity across accessions. To better understand this diversity, the Genomes-to-Fields initiative
37 has produced paired genotypic and phenotypic datasets annually since 2014. Here, we present 15
38 genome assemblies representing 13 maize lines, including founders and testers from the 2022–
39 2023, 2024–2025, and 2026–2027 Genomes-to-Fields evaluation cycles and additional breeding
40 germplasm, together with gene and repeat annotations. Assemblies ranged from 2.19 to 2.38 Gb,
41 with BUSCO completeness of 97.9–99.0%. Excluding the more fragmented LH287 assembly
42 (contig N50 = 5.3 Mb; 90.8% chromosome placement), contig N50 values ranged from 31.1 to
43 195.8 Mb, with 95.9–99.4% of sequence placed on chromosomes. Gene annotation identified
44 approximately 40,000 gene models per genotype, of which 93.6–93.9% were classified as
45 protein-coding, with 95.9–97.3% complete BUSCOs. Comparison with B73 identified 27,606–
46 30,517 one-to-one syntenic orthologs per genotype, representing 69.4–76.8% of B73 genes.
47 Transposable elements occupied 83.6–88.6% of the genomes, with LTR retrotransposons
48 representing the predominant class. Comparative genome alignments revealed three large
49 inversions, including a 12.65-Mb inversion on chromosome 3 present in six of the 13 lines.
50 Together, these resources expand the genomic representation of maize breeding germplasm and
51 provide a foundation for investigating genetic variation underlying agronomic traits.
52


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.15.766802; this version posted November 18, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.15.766802)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


53 **Introduction**
54 Maize ( _Zea mays_ L.) is one of the most widely cultivated crops globally, providing a major source
55 of food, animal feed, and industrial raw materials. Beyond its agricultural significance, maize is a
56 model system for plant genomics and evolution, with a transposable element-rich genome and
57 extensive genetic and structural variation among accessions (Hufford et al., 2021; Schnable et al.,
58 2009). This diversity, shaped by adaptation and breeding, presents opportunities for understanding
59 the genetic basis of complex agronomic traits and their responses to heterogeneous growing
60 environments.
61 Recent advances in long-read sequencing have enabled increasingly complete maize genome
62 assemblies, revealing extensive structural variation that cannot be fully represented by a single
63 reference genome. Comparative analyses of the maize nested association mapping (NAM) founder
64 genomes have highlighted substantial differences in gene content, chromosomal structure, and
65 repetitive sequences (Hufford et al., 2021). Such variation can have important functional
66 consequences. Chromosomal inversions can suppress recombination and maintain linked
67 haplotypes, and have been associated with differences in flowering time, leaf morphology, and
68 yield-related traits in maize and its wild relatives (Crow et al., 2020; Fustier et al., 2019; Romero
69 Navarro et al., 2017). Similarly, heterochromatic knobs, composed of tandem repeats, can
70 influence local recombination and, in certain genetic backgrounds, chromosome transmission,
71 while their repetitive structure presents challenges for genome assembly (Ghaffari et al., 2013;
72 Kanizay et al., 2013; Liu et al., 2020). Despite substantial advances in characterizing maize
73 genomic diversity, much remains to be discovered about the extent of pangenomic variation and
74 its contribution to agronomically important traits across environments.
75 The _Genomes to Fields_ (G2F; &#91;https://www.genomes2fields.org/home/) initiative is a public–&#93;(https://www.genomes2fields.org/home/)
76 private partnership established in 2013 to advance understanding of how maize genetic variation
77 influences phenotype across diverse environments. Through multidisciplinary collaboration, G2F
78 integrates publicly available genomic, phenotypic, and environmental data with computational
79 modeling to characterize genotype-by-environment (G×E) interactions, improve predictions of
80 agronomically important traits, and inform crop improvement (Chen et al., 2026; Lima et al.,
81 2023).
82 Here, we present genomic resources comprising long-read sequencing data, 15 genome assemblies
83 representing 13 maize lines, and associated gene and repeat annotations. These resources
84 encompass founder, tester, and additional breeding germplasm from the 2022–2023, 2024–2025,
85 and 2026–2027 G2F evaluation cycles. Comparative analyses identified three large chromosomal
86 inversions, assessed gene model correspondence with B73, and characterized transposable element
87 content across these genomes. Together, these resources support future investigations of maize
88 genomic diversity and genotype–phenotype relationships within G2F.
89

90 **Methods and materials**

91 **Germplasm selection**
92 Four stiff-stalk inbred lines (‘3AZA1’, ‘FBLL’, ‘PHB47’, and ‘PHBW8’) were crossed with four
93 stiff-stalk-derived Germplasm Enhancement of Maize (GEM) lines: ‘GEMS-0061’, ‘GEMS94 0113’, ‘GEMS-0219’, and ‘GEMS-0227’ (Rogers et al., 2022). Doubled-haploid (DH) lines were
95 derived from the resulting F1 progeny and subsequently testcrossed with the ex-Plant Variety
96 Protection (ex-PVP) inbred lines ‘PHP02’ and ‘LH287’ for the 2024–2025 evaluation cycle
97 (Figure 1; &#91;https://www.genomes2fields.org/about/project-overview/). Three additional lines&#93;(https://www.genomes2fields.org/about/project-overview/)
98 represent other evaluation cycles: ‘LH244’, a stiff-stalk tester used in 2022–2023, and ‘PH09B’


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.15.766802; this version posted November 18, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.15.766802)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


99 and ‘PH2MW’, ex-PVP inbreds included in the 2026–2027 evaluation cycle. The pedigree, origin,
100 and distinguishing characteristics of each line are described below, based on records from the
101 USDA National Plant Germplasm System (GRIN-Global), Plant Variety Protection (PVP)
102 documentation, and published germplasm reports (National Plant Germplasm System, 2025;
103 White et al., 2020).
104 **3AZA1:** Former DEKALB Genetics Corporation PVP (PVP 9600027) maintained by the USDA
105 National Plant Germplasm System (GRIN-Global) under PI 592379. This is a highly inbred stiff
106 stalk pedigree line developed from proprietary germplasm incorporating several public inbreds.
107 3AZA1 was developed by crossing AQA3 X QFAR, with AQA3 derived from A632, CM105,
108 &#91;B14, and LH23, and QFAR derived from B14, B37, and LH23 (USDA GRIN-Global, PI 592379).&#93;(https://npgsweb.ars-grin.gov/gringlobal/accessiondetail?id=1502955)
109 Its characteristics include yellow semident kernels, red cobs, and approximately 16 cm ears.
110 **FBLL:** Former DEKALB Plant Genetics PVP-associated inbred (PVP 9100034) maintained by
111 GRIN-Global under PI 546481. This is a highly inbred, yellow dent line with a breeding
112 background related to B73-type germplasm. FBLL was developed by crossing the proprietary
113 DEKALB lines 5B2C-A X PB80 in 1982, followed by multiple generations of self-pollination and
114 &#91;selection, USDA Plant Inventory No. 200 (Pfund, 2002). Its characteristics include yellow dent&#93;(https://www.ars-grin.gov/npgs/pi_books/scans/pi200pt1.pdf)
115 kernels, red cobs, approximately 14.4 cm ears, and an average of 17 kernel rows per ear.
116 **GEMS-0061:** Germplasm Enhancement of Maize (GEM) breeding line maintained by GRIN117 Global under PI 685801. This is a partially inbred, exotic-derived line developed through the
118 USDA-ARS GEM project using South American maize germplasm. GEMS-0061 was developed
119 from the Argentine accession ARZM 16 026 (PI 516027), originating from Mendoza, crossed with
120 the anonymous stiff stalk GEM donor S17, with the selected pedigree AR16026:S17-10-1-B-B
121 &#91;(USDA GRIN-Global, PI 685801). Its characteristics include predominantly yellow to orange-red&#93;(https://npgsweb.ars-grin.gov/gringlobal/accessiondetail?id=1645902)
122 kernels exhibiting semident and semiflint morphology, white cobs, and approximately 15.5 cm
123 ears.
124 **GEMS-0113:** Germplasm Enhancement of Maize (GEM) breeding line maintained by GRIN125 Global under PI 651540. This is a partially inbred, exotic-derived line released through the USDA126 ARS GEM project in 2005. GEMS-0113 was developed from the Mexican accession CHIS775 (PI
127 515164) and the S1911b breeding donor, with the selected pedigree CHIS775:S1911b-120-001128 &#91;B-B-SIB (USDA GRIN-Global, PI 651540). Its characteristics include yellow semident kernels,&#93;(https://npgsweb.ars-grin.gov/gringlobal/accessiondetail?id=1740658)
129 red cobs, regularly arranged kernel rows, and approximately 14–15 cm ears.
130 **GEMS-0219:** Germplasm Enhancement of Maize (GEM) breeding line maintained by GRIN131 Global under PI 685914. This is a partially inbred, exotic-derived line developed through the
132 USDA-ARS GEM project and released in 2011. GEMS-0219 was developed from the Cuban
133 maize accession CUBA164 (PI 489361) and GEM breeding line GEMS-0002 (PI 632413), with
134 &#91;the selected pedigree (CUBA164:S2012-459-001-B/GEMS-0002)-B-B-031-B-B-SIB (USDA&#93;(https://npgsweb.ars-grin.gov/gringlobal/accessiondetail?id=1914784)
135 &#91;GRIN-Global, PI 685914). Its characteristics include yellow semident kernels, white cobs,&#93;(https://npgsweb.ars-grin.gov/gringlobal/accessiondetail?id=1914784)
136 approximately 13–14.5 cm ears, and plant heights of approximately 242 cm in a USDA 2016
137 evaluation.
138 **GEMS-0227:** Germplasm Enhancement of Maize (GEM) breeding line maintained by GRIN139 Global under PI 685920. This is a partially inbred, exotic-derived line developed through the
140 USDA-ARS GEM project and released in 2012. GEMS-0227 was developed from the tropical
141 maize accession NEI9004 (Ames 29045), associated with Suwan germplasm, with the selected
142 pedigree NEI9004:S2809-B-022 &#91;(USDA GRIN-Global, PI 685920). Its characteristics include&#93;(https://npgsweb.ars-grin.gov/gringlobal/accessiondetail?id=1939959)
143 yellow semident kernels, red cobs, approximately 14 kernel rows, and 15.5 cm ears.


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.15.766802; this version posted November 18, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.15.766802)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


144 **LH244:** Former Holden's Foundation Seeds inbred (PVP 200000122) maintained by GRIN-Global
145 under PI 612589. This is a highly inbred stiff stalk line developed from Holden's breeding
146 germplasm and used as a tester in the Genomes to Fields 2022/2023 evaluation cycle. LH244 was
147 developed by crossing LH197 X (LH197 X LH199), representing a backcross toward LH197
148 &#91;(USDA GRIN-Global, PI 612589). Its characteristics include yellow dent kernels, red cobs,&#93;(https://npgsweb.ars-grin.gov/gringlobal/accessiondetail?id=1594201)
149 approximately 14–16 kernel rows, and 15.5–17 cm ears. LH244 is also notable for its high
150 transformation efficiency and extensive genomic similarity to B73 (Tan et al., 2026).
151 **LH287:** Former Monsanto inbred line (PVP 200000121, maintained by GRIN-Global as PI
152 612588) developed via a cross between LH212Ht X LH185 (Foley, 2001). LH287 is a yellow dent,
153 non-stiff stalk line that is widely used as a tester in hybrids, including in the Genomes to Fields
154 initiative &#91;(https://www.genomes2fields.org/about/project-overview/). Characteristics of LH287&#93;(https://www.genomes2fields.org/about/project-overview/)
155 include pollen shed at 70 days post emergence, yellow-orange dent kernels, white cobs and 16 cm
156 ears.
157 **PH09B:** Former Pioneer Hi-Bred PVP (PVP 9700218) maintained by the USDA National Plant
158 Germplasm System (GRIN-Global) under PI 597331. This is a high-yielding, highly inbred stiff
159 stalk line frequently used in commercial hybrids with PH2MW. This accession was developed by
160 crossing PHP38 X PHHB9, PVP Certificate (USDA Plant Variety Protection Office, 2000), and
161 its characteristics include pollen shed at approximately 74 days post emergence, yellow dent
162 kernels, red cobs, and 17-19 cm ears.
163 **PH2MW:** Former Pioneer Hi-Bred PVP (PVP 9900382) maintained by GRIN-Global as PI
164 612160. This is a non-stiff-stalk line historically used as a pollen parent in hybrid production.
165 PH2MW was developed by crossing PHGG6 X PHAW6, PVP Certificate (USDA Plant Variety
166 Protection Office, 2001), and its characteristics include pollen shed at 76 days post emergence,
167 yellow dent kernels, red cobs, and 18 cm ears.
168 **PHB47:** Former Pioneer Hi-Bred PVP (PVP 8300141, registered as B47) maintained by GRIN169 Global under PI 601009. This is a highly inbred stiff stalk line derived from public maize breeding
170 germplasm. PHB47 was developed by crossing B37 X SD105, with B37 originating from Iowa
171 &#91;Stiff Stalk Synthetic and SD105 representing an early public South Dakota inbred (USDA GRIN-&#93;(https://npgsweb.ars-grin.gov/gringlobal/accessiondetail?id=1157410)
172 &#91;Global, PI 601009). Its characteristics include yellow kernels exhibiting semident to semiflint&#93;(https://npgsweb.ars-grin.gov/gringlobal/accessiondetail?id=1157410)
173 morphology, red cobs, and approximately 14–16 cm ears in multiple USDA evaluations.
174 **PHBW8:** Former Pioneer Hi-Bred PVP (PVP 9200079) maintained by GRIN-Global under PI
175 559936. This is a highly inbred stiff stalk line developed from Pioneer breeding germplasm.
176 PHBW8 was developed by crossing PHJ40 X PHW52 &#91;(USDA GRIN-Global, PI 559936). Its&#93;(https://npgsweb.ars-grin.gov/gringlobal/accessiondetail?id=1454908)
177 characteristics include yellow semident kernels, red cobs, approximately 14 kernel rows, and ears
178 measuring 18 cm in a USDA 2014 evaluation.
179 **PHP02:** Former Pioneer Hi-Bred PVP (PVP 8800212) maintained by GRIN-Global under PI
180 601570. This is a highly inbred line used as a tester in commercial maize breeding and in the
181 &#91;Genomes to Fields initiative. PHP02 was developed by crossing PHG44 X PHG29 (USDA GRIN-&#93;(https://npgsweb.ars-grin.gov/gringlobal/accessiondetail?id=1185648)
182 &#91;Global, PI 601570). Its characteristics include yellow dent kernels, red cobs, and approximately&#93;(https://npgsweb.ars-grin.gov/gringlobal/accessiondetail?id=1185648)
183 14.5 cm ears.

184 **Genome sequencing and assembly**
185 PacBio HiFi reads were generated on the Revio system and processed on-instrument using ccs
186 v8.0.1, trim v1.1.0, and lima v2.10.0. HiFi BAM files were converted to FASTQ using bam2fastq
187 (pbtk v3.5.0; &#91;https://github.com/PacificBiosciences/pbtk/tree/v3.5.0).&#93;(https://github.com/PacificBiosciences/pbtk/tree/v3.5.0) Oxford Nanopore
188 Technologies (ONT) reads (R10.4.1 chemistry, PromethION 2 Solo) were basecalled using
189 MinKNOW with the HAC v4.2.0 model (SUP v4.2.0 for PHP02). Reads passing MinKNOW


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.15.766802; this version posted November 18, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.15.766802)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


190 quality filtering and ≥50 kb in length were selected for hybrid assembly using SeqKit v2.13.0
191 (Shen et al., 2016).
192 Read-set identities were assessed using k-mer fingerprinting ( _k_ = 21 and 31) with sourmash v4.9.3
193 (Irber et al., 2024). Potential sample discordance was further investigated by examining haplotype
194 structure across individual SMRT cells, allele concordance between sequencing technologies, and
195 the alignment of reads to their corresponding assemblies.
196 Genome assemblies were generated using hifiasm v0.25.0-r726 (Cheng et al., 2021, 2024),
197 incorporating HiFi and ONT ultra-long reads where appropriate. Assembly configurations were
198 selected based on read concordance, haplotype structure, assembly completeness, sequence
199 duplication, and structural consistency. For six lines (FBLL, PHBW8, PHB47, 3AZA1, PHP02,
200 and GEMS-0227), unpurged primary contigs (-l0) were retained. For GEMS-0061 and GEMS201 0113, partially phased assemblies were generated using haplotig purging (-l3) with adjusted
202 primary coverage parameters (--pri-range 24 and 23, respectively), and both haplotype partitions
203 were retained. Evidence of mixed biological material in GEMS-0219 prompted assembly from
204 HiFi reads originating from a single SMRT cell (bc2041), using -l3 --pri-range 16 without ONT
205 reads and retaining only the primary assembly.
206 **Assembly quality control**
207 Assembly contiguity statistics were calculated using SeqKit. Gene-space completeness was
208 evaluated using BUSCO v6.1.0 (Tegenfeldt et al., 2025) in genome mode with miniprot and the
209 poales_odb12.2 lineage dataset (6,188 orthologs). B73 RefGen_v5 (Hufford et al., 2021) was
210 evaluated using the same procedure for comparison. Consensus quality values (QV) and k-mer
211 completeness were estimated using Merqury v1.3 (Rhie et al., 2020) with _k_ = 21 databases
212 generated from the corresponding HiFi reads.
213 Assembly integrity was evaluated using read alignments, assembly graphs, and synteny with B73
214 RefGen_v5. HiFi reads were aligned to their respective assemblies using minimap2 v2.30; map215 hifi (Li, 2018), with alignments processed using SAMtools v1.20 (Danecek et al., 2021) to assess
216 sequence coverage and structural support. Gene-anchor synteny relative to B73 was evaluated
217 using AnchorWave v1.3.1 (Song et al., 2022). Structural differences from B73 were not considered
218 assembly errors without additional supporting evidence. Six unsupported interchromosomal joins
219 were resolved by sequence-preserving contig splits, including two contradicted by read evidence
220 and four associated with unsupported joins within (AGT)n repeat arrays.
221 All 11 contig assemblies were screened for foreign-organism contamination with NCBI FCS-GX
222 v0.5.5 (Astashyn et al., 2024) against the GX database (build 2023-01-24) using taxonomy ID
223 4577 (Zea mays), and for adaptor and vector sequence with FCS-adaptor v0.5.5 in eukaryotic mode
224 (--euk).
225 **Reference-guided scaffolding**
226 Contigs were then assigned to chromosomes, ordered, and oriented relative to B73 RefGen_v5
227 using ALLMAPS JCVI v1.6.6 (Tang et al., 2015). A maize pan-genome anchor-marker set
228 comprising 100-bp sequences was mapped to B73 to establish reference positions. Markers were
229 subsequently aligned to the assembled contigs using HISAT2 v2.2.1 (Kim et al., 2019), retaining
230 uniquely mapped alignments (MAPQ 60; edit distance ≤5). Contigs of at least 100 kb containing
231 at least 10 unique markers at five distinct reference positions were eligible for placement.
232 Scaffolds were constructed using ALLMAPS, with a maximum of 100 evenly distributed markers
233 per contig. Adjacent contigs were joined using 100-N gaps of unknown physical length, without
234 trimming or splitting contig sequences. Scaffold joins were evaluated using HiFi read alignments,
235 assembly-graph connectivity, and ONT reads where available from the same biological sample.


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.15.766802; this version posted November 18, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.15.766802)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


236 ONT reads from GEMS-0061 and GEMS-0113 were excluded from join validation because of
237 uncertain biological concordance with the HiFi samples. Joins were classified according to direct
238 sequence support or reference-based placement, and conflicting evidence was individually
239 reviewed.
240 Contigs lacking sufficiently supported chromosome assignments were retained as unplaced
241 sequences. Scaffold sequences were validated against their AGP representations to confirm that
242 each contig was incorporated completely and without duplication. Because chromosome
243 assignments and contig order were inferred using B73, the resulting scaffolds represent reference244 guided chromosome organization rather than independently resolved chromosome-scale
245 assemblies.
246 Telomeric repeat arrays were identified with tidk v0.3.2; motif TTTAGGG; -w 10000 --min-length
247 200 --max-offset 1000 (Brown et al., 2025), and chromosome ends were classified from these
248 arrays. The Mo17 CAU-2.0 telomere-to-telomere (T2T) assembly and B73 RefGen_v5 served as
249 controls.
250 The assembly pipeline was implemented as a containerized Snakemake workflow v9.26.1 (Mölder
251 et al., 2021), with accompanying documentation and AI-agent instructions to facilitate
252 reproducible execution and maintenance.
253 **Genome annotation**
254 Genomes were annotated using Helixer 0.3.3, which leverages a cross-species deep learning model
255 for accurate _ab initio_ gene structure prediction (Holst et al., 2026). The model identifies genomic
256 regions corresponding to the untranslated regions (UTRs), coding sequences (CDSs), and introns
257 of genes. For each species, one of four predefined lineage-specific models (vertebrate,
258 invertebrate, land plant, or fungi) is selected based on its taxonomic classification. Here, for maize,
259 the land plant lineage model was used.
260 **Gene model annotation validation**
261 The completeness of gene annotations was evaluated with Benchmarking Universal Single-Copy
262 Orthologs (BUSCO v6.1.0) in protein mode, using the Poales lineage dataset (poales_odb12),
263 which contains 6,282 conserved single-copy orthologs. For each of the genomes, the predicted
264 proteins were searched against the lineage dataset, and the numbers of complete (single-copy and
265 duplicated), fragmented, and missing BUSCOs were recorded. To assess whether the predicted
266 gene models are likely to encode proteins, every Helixer-predicted gene model was scored with
267 Protein Sequence Assessment Using a Reference ORF Network (PSAURON v1.1.3) (Sommer et
268 al., 2025). PSAURON is a machine-learning model trained on annotated protein-coding sequences
269 to estimate the probability that a CDS encodes a protein, both in its annotated reading frame (in270 frame score) and in the five alternative frames (out-of-frame scores). Extracted CDS sequences
271 were run on PSAURON using default settings, and gene models with an in-frame score ≥0.5 and
272 out-of-frame scores ≤0.5 were classified as true protein-coding sequences.
273 **Orthologous genes between each genome and B73**
274 Syntenic genes between each of the genomes and B73 (RefGen_v5) were identified with MCScan
275 (JCVI utility libraries v1.6.8) (Tang et al., 2024). B73 gene model and the gene models of each
276 genotype were aligned with LAST (Kiełbasa et al., 2011), and the parameters “--cscore=.99 -277 quota=1:1” were used to find 1-to-1 syntenic gene relationships. A similar approach was used to
278 identify syntenic genes between A188 and B73 (Yao et al., 2026). To validate this approach,
279 Liftoff was also used to lift over all genomes to the B73 (RefGen_v5) coordinate system to identify
280 genes that matched B73 genes. The resulting lifted GFF3 files were processed to extract CDS
281 features, and the overlapping CDS regions between the B73 reference genes and transferred gene


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.15.766802; this version posted November 18, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.15.766802)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


282 models from each of the genomes were identified using BEDTools intersect (Quinlan &amp; Hall,
283 2010). The cross-referenced, corresponding gene identifiers were extracted from the resulting
284 intersection files using the Pandas package in Python (V3.14.10). For both analyses, a candidate
285 orthologous gene matrix was constructed using B73 genes as the reference with 1-to-many gene
286 relationships for liftoff and 1-to-1 relationship for MCScan.
287 **Transposable element annotation**
288 Transposable elements (TEs) were annotated across the genomes using EDTA v2.3.0 (Ou et al.,
289 2019) with maize-specific settings (--species Maize). The pipeline identified different TE classes
290 using LTRharvest (Ellinghaus et al., 2008), LTR_FINDER (Xu &amp; Wang, 2007), TIR-Learner (Su
291 et al., 2019), HelitronScanner (Xiong et al., 2014) AnnoSINE2 v2.0.9, and the EDTA LINE
292 module. The TE libraries were combined and curated using EDTA, accompanied by de novo
293 predictions from RepeatModeler v2.0.7 (Flynn et al., 2020), and used by RepeatMasker v4.2.3
294 &#91;(http://www.repeatmasker.org) to annotate intact and fragmented TE copies. The genomic&#93;(http://www.repeatmasker.org/)
295 coverage of each TE class was then calculated across genomes.
296 **AI usage**
297 Claude Code (Opus 5.5; v2.1.318; Anthropic) was used to generate and audit code. ChatGPT
298 (GPT-6; OpenAI) was used for reviewing and editing writing and to audit Claude Code output.
299

300 **Results and discussion**

301 **Sequencing and genome assembly**
302 Genome assemblies for thirteen maize lines were generated using accession-specific sequencing
303 and assembly strategies, including PacBio HiFi and Oxford Nanopore (ONT) long-read
304 sequencing. Among twelve lines with available sequencing records, HiFi inputs ranged from 57 to
305 150 Gb (25–65× coverage), while ONT reads ≥50 kb contributed 0.03–1.95 Gb (less than 1×
306 coverage; Supplementary Table 1). Differences in residual heterozygosity and sequencing data
307 quality required accession-specific assembly methods, resulting in 11 primary assemblies and two
308 partially phased assembly pairs.
309 The 11 primary assemblies ranged from 2.19 to 2.38 Gb, with contig N50 values of 94.9–195.8
310 Mb for 10 assemblies, while LH287 was substantially more fragmented (N50 = 5.3 Mb). The
311 partially phased haplotype partitions (2.24–2.30 Gb) had contig N50 values of 31.1–85.8 Mb
312 (Table 1). Assembly completeness was high, with 97.9–99.0% of conserved BUSCO genes
313 recovered and estimated consensus QV values of 67.5–76.0. Primary assemblies recovered 95.8–
314 98.7% of read-derived k-mers, while the partially phased haplotype pairs collectively recovered
315 approximately 98%. Across all 15 assemblies, three contigs spanned more than one chromosome
316 (two unresolved terminal joins in GEMS-0227 and one misjoin in PH2MW), telomeric repeat
317 arrays ≥1 kb were detected at 249 of 300 chromosome ends (83.0%), and 12 chromosomes were
318 assembled as single gap-free contigs with telomeric arrays at both ends. To facilitate downstream
319 comparative analyses, assemblies were organized into chromosome-assigned scaffolds using
320 accession-specific approaches, placing 95.9–99.4% of assembled sequence (90.8% for LH287)
321 into scaffolds assigned to the 10 maize chromosomes. Because some intercontig joins were
322 supported by reference order alone, these scaffolds should be interpreted as chromosome
323 assignments rather than uniformly contiguous chromosome-level assemblies.
324 **Genomic structural variation**
325 Comparative alignments with B73 RefGen_v5 revealed three well-supported large inversions
326 across the 13 maize lines (Figure 2). Inversion arrangements were identified from assembled


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.15.766802; this version posted November 18, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.15.766802)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


327 sequences spanning both breakpoints or, for the fragmented LH287 assembly, from long reads
328 supporting both inversion junctions.
329 The largest inversion spanned 12.65 Mb on chromosome 3 (97.6–110.3 Mb) and was present in
330 six of 13 lines (GEMS-0113, GEMS-0219, LH244, LH287, PH2MW and PHBW8). Breakpoints
331 were resolved to base-pair precision and were identical among assembled carriers. In five lines,
332 17–39 HiFi or ONT reads per junction supported the inverted arrangement, with no qualifying
333 reads supporting the B73 orientation; LH244 was supported by assembly alignments alone. Both
334 junction sequences were also present in the genome assemblies of NAM founders P39 and Oh43
335 (Hufford et al., 2021), and corresponded to inversion SV3 previously reported in LH244 (Tan et
336 al., 2026).
337 A second inversion spanning 2.05 Mb on chromosome 9 (84.9–86.9 Mb) was identified in seven
338 of 13 lines (3AZA1, GEMS-0113, GEMS-0227, LH287, PH2MW, PHBW8 and PHP02), with 23–
339 54 orientation-switching long reads supporting each junction in all seven lines. A 6.98-Mb
340 inversion on chromosome 3 (36.9–43.8 Mb) was found only in PHP02 among the 12 lines that
341 could be assessed, with more limited read support at both junctions; the same junction sequences
342 were present in the NAM founder CML228 assembly (Hufford et al., 2021). The two partially
343 phased partitions of GEMS-0061 and GEMS-0113 did not disagree at any of the three loci. Overall,
344 nine of the 13 lines carried at least one of these inversion polymorphisms, demonstrating the
345 established structural variation known to exist across maize lines.

346 **Gene annotation and completeness**
347 Similar numbers of genes were annotated in each genome. Helixer predicted between 39,868
348 (GEMS-0113 hap2) and 41,074 (PH2MW) gene models across the genotypes (Table 2). Each gene
349 model had a single transcript, so mRNA and gene counts were identical. The average CDS lengths
350 were ~1,139 bp across the genotypes, with the longest average found in line PHBW8 at 1,145 bp
351 and the shortest average found in line GEMS-0061 hap2 at 1,134 bp. Overall, gene feature
352 statistics, including mean CDS length, exon and intron lengths, protein length, and exon number,
353 were highly consistent across the genotypes (Figure 3; Supplementary Table 2). BUSCO analysis
354 reported 95.9–97.3% complete BUSCOs across the genotypes. Complete BUSCOs comprised
355 70.1–72.3% single-copy and 24.8–27.0% duplicated genes (Table 2; Supplementary Table 3).
356 PSAURON, on the other hand, classified 93.6–93.9% of the gene models in each genotype as
357 protein-coding (Table 2), with a mean in-frame score of 0.906–0.909 (Supplementary Table 4).
358 Low-scoring gene models (in-frame score &lt; 0.5) were more common among genes with shorter
359 CDS lengths, and their count decreased as CDS length increased. Single-exon gene models also
360 had a higher proportion of low scores than multi-exon models across all genotypes (Supplementary
361 Table 4).
362 **Annotated gene model comparisons with B73**
363 The 1-to-1 syntenic orthologs between B73 and each genotype identified 27,606 - 30,517 pairs,
364 where 69.4–76.8% of B73 genes had an ortholog in any one genotype (Figure 4 &amp; Table 3). FBLL
365 (76.8%) and LH244 (76.1%) had the most orthologs, while LH287 (69.4%), PH2MW (69.9%),
366 and PHP02 (70.2%) had the lowest number of orthologs. Between 10,160 and 13,277 genes per
367 genotype had no one-to-one syntenic B73 partner, potentially because: 1) the 1-to-1 requirement
368 leaves out tandem and other duplicate copies; 2) some genes may have moved out of syntenic
369 position; 3) genes are derived from transposable elements; and 4) gene models were not present in


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.15.766802; this version posted November 18, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.15.766802)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


370 the B73 annotation. As an independent check, we used Liftoff and matched each gene model to
371 B73 genes by CDS overlap. Matching genes assigned by Liftoff agreed with MCScan 94.5–97.4%
372 of the time (Supplementary Table 5). Liftoff matched a higher proportion of genes to B73 than
373 MCScan (76.3–79.8%) per genotype and left fewer genes unmatched (8,730–9,950). This
374 difference is likely because Liftoff does not require one-to-one matches; therefore, several genes
375 can be assigned to the same B73 gene. For instance, Liftoff mapped 17–29% of genes lacking a
376 syntenic partner to B73, suggesting that many of these genes are not absent from B73 but may
377 potentially represent duplicated copies excluded by MCScan’s one-to-one requirement. Together,
378 the two approaches characterize how annotated genes correspond to existing B73 annotations.
379 MCScan identifies strict one-to-one syntenic orthologs, while Liftoff shows where a greater
380 proportion of genes map, including duplicates. Supplementary Tables 6 and 7 provide the
381 corresponding gene models for each B73 gene across the genotypes, as identified by MCScan (1382 to-1 relationships) and Liftoff (1-to-many relationships), respectively.

383 **TE Annotation**

384 TEs occupied 1,712–2,040 Mb of the nuclear genome, corresponding to 83.6–88.6% of the
385 genome (Supplementary Table 8), consistent with the high repeat content of maize genomes
386 (Hufford et al., 2021). LTR retrotransposons were the most abundant class of TEs, occupying
387 76.9–79.6% of the nuclear genome. Structural annotation identified 54,183–59,741 intact LTR
388 retrotransposons per genome, of which 52% were Gypsy, 38% Copia, and 11% unclassified. Intact
389 TIR elements and intact Helitrons were collapsed into non-redundant libraries for each genome,
390 with an average of approximately 10,000 LTR, 5,000 TIR, 2,000 Helitrons, and 16 LINE families
391 per genome (Supplementary Table 8). Helitron detection failed in LH287, PH09B, and PH2MW;
392 therefore, Helitron content was not reported for these genotypes, resulting in a slight

393 underestimation of their total TE content.

394 **Conclusions**

395 We present high-quality genome assemblies and annotations for diverse maize germplasm.
396 Comparative analyses revealed shared and accession-specific structural variation, including large
397 chromosomal inversions, highlighting genomic diversity not captured by a single reference
398 genome. Together, these resources provide a foundation for investigating the inheritance and
399 functional consequences of genomic variation and for improving our understanding of genotype–
400 phenotype relationships across environments in maize.

401


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.15.766802; this version posted November 18, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.15.766802)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


402 **Tables and figures**
403 **Table 1: Assembly statistics and quality metrics for thirteen maize lines**



**Chr.**

**placed**
**(%)**



**Size**
**Assembly** **Contigs** **&#91;N50 &#93;**
**(Gb)** **(Mb)**



**BUSCO C**
**QV**
**(D) (%)**



**k-mer**

**comp.**
**(%)**



**Telomeres**

**≥1 kb**



3AZA1 2.326 185 138.3 98.9 (24.9) 74.2 98.67 99.0 19/20

FBLL 2.247 187 112.3 98.9 (24.9) 75.2 98.45 99.0 19/20

GEMS-0061
2.300 228 31.1 98.3 (25.7) 74.9 86.78 98.2 15/20
H1


GEMS-0061
2.246 187 41.2 98.3 (25.7) 74.3 86.13 98.9 15/20
H2


GEMS-0113
2.263 194 85.8 98.1 (25.7) 76.0 91.26 98.3 19/20
H1


GEMS-0113
2.236 176 55.3 97.9 (24.9) 74.6 90.83 99.1 19/20
H2


GEMS-0219 2.236 168 103.7 99.0 (25.7) 73.9 97.24 98.9 20/20

GEMS-0227 2.378 935 94.9 98.9 (26.8) 71.0 98.57 95.9 19/20

LH244 2.282 144 195.8 99.0 (25.0) NA NA 99.4 19/20

LH287 2.231 3,691 5.3 98.7 (24.9) 71.7 98.54 90.8 11/20

PH09B 2.291 232 164.5 99.0 (25.3) 67.5 96.33 99.3 11/20

PH2MW 2.216 219 139.6 98.9 (25.5) 72.3 95.75 99.3 9/20

PHB47 2.188 183 114.7 98.9 (24.9) 74.1 98.72 99.0 18/20

PHBW8 2.249 268 121.1 98.9 (25.0) 73.4 98.34 98.9 19/20

PHP02 2.228 253 120.8 98.9 (25.1) 73.8 98.48 98.4 17/20

404
405 _C, complete BUSCOs; D, duplicated BUSCOs (poales_odb12.2); QV, consensus quality value;_
406 _H1/H2, partially phased assembly partitions; NA, not available. Telomeres indicate the number_
407 _of evaluated chromosome ends with a telomeric repeat array ≥1 kb within 1 kb of the terminal_
408 _contig end. For GEMS-0113 H2, chromosome 4 is represented by two scaffolds, and only the two_
409 _outer ends were counted. Chromosome placement reflects B73-guided scaffolding, except for_
410 _LH244 (provider-supplied scaffolds; methods not reported), and does not necessarily indicate_
411 _contiguous chromosome-scale assemblies. Combined k-mer completeness of the two partitions_
412 _was 98.02% for GEMS-0061 and 97.96% for GEMS-0113. QV and k-mer completeness for_
413 _LH287, PH09B and PH2MW were estimated using collaborator-provided reads; these metrics_
414 _were unavailable for LH244 because reads were not available for evaluation. GEMS-0227 is an_
415 _unpurged primary assembly; GEMS-0219 represents a single-plant primary assembly._
416

417


418


419


420


421


422


423 **Table 2: Summary statistics for the gene features predicted by Helixer on each genome assembly**





















|Predicted&lt;br&gt;features|FBLL|PHBW8|PHB47|3AZA1|PHP02|GEMS&lt;br&gt;-&lt;br&gt;0227|GEMS&lt;br&gt;-&lt;br&gt;0219|GEMS&lt;br&gt;-&lt;br&gt;0061&lt;br&gt;hap1|GEMS&lt;br&gt;-&lt;br&gt;0061&lt;br&gt;hap2|GEMS&lt;br&gt;-&lt;br&gt;0113&lt;br&gt;hap1|GEMS&lt;br&gt;-&lt;br&gt;0113&lt;br&gt;hap2|LH287|PH09B|PH2MW|LH244|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
|gene|40,677|40,563|40,465|40,594|40,705|40,723|40,777|40,873|40,724|40,741|39,868|39,984|40,956|41,074|40,585|
|mRNA|40,677|40,563|40,465|40,594|40,705|40,723|40,777|40,873|40,724|40,741|39,868|39,984|40,956|41,074|40,585|
|CDS|217,631|218,051|216,632|217,606|218,523|218,094|218,388|218,153|217,338|218,650|214,651|213,953|218,850|219,502|217,147|
|exon|239,164|239,515|238,113|239,128|239,959|239,760|240,033|239,640|238,852|240,193|235,602|234,999|240,249|241,292|238,687|
|5′ UTR|51,142|51,048|50,967|51,125|51,357|51,324|51,266|51,481|51,356|51,186|50,096|50,249|51,496|51,825|51,062|
|3′ UTR|50,896|50,667|50,567|50,702|50,627|50,892|51,060|50,905|50,785|50,950|49,699|49,901|50,926|51,241|50,776|
|Complete&lt;br&gt;BUSCOs&lt;br&gt;(%)|97.3|97.1|97.2|97.0|97.1|97.1|97.0|96.6|96.2|96.3|95.9|96.9|97.2|97.0|97.0|
|PSAURON&lt;br&gt;Score (%)|93.9|93.9|93.9|93.7|93.9|93.9|93.7|93.6|93.7|93.7|93.8|93.7|93.9|93.8|93.8|


424


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.15.766802; this version posted November 18, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.15.766802)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


425 **Table 3: One-to-one syntenic orthologs between B73 and each genotype**

















426


|Genotype|Genes&lt;br&gt;per&lt;br&gt;Genotype|One-to-one&lt;br&gt;B73gene&lt;br&gt;match|%B73&lt;br&gt;gene&lt;br&gt;match|Gene&lt;br&gt;without&lt;br&gt;B73match|Synteny&lt;br&gt;blocks|Lifted&lt;br&gt;pairs|
|---|---|---|---|---|---|---|
|FBLL|40677|30517|76.8|10160|18|3382|
|PHBW8|40563|29208|73.5|11355|20|2874|
|PHB47|40465|29085|73.2|11380|14|2828|
|3AZA1|40594|29191|73.4|11403|15|2889|
|PHP02|40705|27890|70.2|12815|20|2631|
|GEMS-0227|40723|29308|73.7|11415|20|3017|
|GEMS-0219|40777|28936|72.8|11841|21|2995|
|GEMS-0061_hap1|40873|29157|73.3|11716|23|3068|
|GEMS-0061_hap2|40724|28353|71.3|12371|32|2886|
|GEMS-0113_hap1|40741|29182|73.4|11559|25|3173|
|GEMS-0113_hap2|39868|29346|73.8|10522|28|3197|
|LH287|39984|27606|69.4|12378|25|2572|
|PH09B|40956|29726|74.8|11230|22|3105|
|PH2MW|41074|27797|69.9|13277|31|2623|
|LH244|40585|30259|76.1|10326|17|3238|


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.15.766802; this version posted November 18, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.15.766802)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


427


428 **Figure 1. Breeding scheme and germplasm selection for the Genomes-to-Fields maize**
429 **population.** Four stiff-stalk inbred lines (3AZA1, FBLL, PHB47, and PHBW8) were crossed
430 with four stiff-stalk-derived Germplasm Enhancement of Maize (GEM) lines (GEMS-0061,
431 GEMS-0113, GEMS-0219, and GEMS-0227), generating 16 F1 families. Doubled-haploid (DH)
432 lines derived from these crosses were testcrossed with PHP02 and LH287 to produce hybrids for
433 the 2024–2025 field evaluation cycle. Three additional assembled lines represent other
434 evaluation cycles: LH244 (2022–2023) and PH09B and PH2MW (2026–2027).


435


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.15.766802; this version posted November 18, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.15.766802)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


436


437 **Figure 2: Chromosomal inversions in G2F 2026 maize assemblies.** B73 v5 ideograms
438 indicate two regions on chromosome 3 (36.853–43.829 and 97.615–110.265 Mb) and one on
439 chromosome 9 (84.885–86.938 Mb), enlarged below (dashed lines). Assembly contigs are shown
440 at their original lengths, aligned to B73. Blue (chromosome 3) and green (chromosome 9)
441 indicate sequences in B73 orientation, magenta indicates sequences aligned in reverse orientation
442 relative to B73, and grey indicates regions without ordered B73 alignments. Ribbons connect
443 corresponding sequences, with crossings indicating orientation differences. Black bars mark
444 contig ends; 'rc' indicates reverse-complemented contigs; and hap1/hap2 denote the two partially
445 phased partitions of GEMS-0061 and GEMS-0113. LH287 is not shown because its assembly is
446 fragmented at these loci.


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.15.766802; this version posted November 18, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.15.766802)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


447


448
449 **Figure 3: Comparison of gene model characteristics across 13 maize genotypes.**
450 Distributions of gene length (kb), coding sequence (CDS) length (bp), exon number per gene,
451 and predicted protein length (amino acids) across annotated gene models. Boxplots show the
452 median, interquartile range, and whiskers. Gene model characteristics were broadly consistent
453 across genotypes.
454


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.15.766802; this version posted November 18, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.15.766802)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


455


456
457 **Figure 4** : **A classification of annotated genes by their similarity to existing B73 annotations**
458 **(AGPv5).** The core genes (green) represent genes shared across all 13 lines; near-core (blue) genes
459 are shared in 11-12 lines; dispensable genes (teal) represent genes shared in 2 -10 genotypes;
460 private genes (yellow) represent genes found in one line; and purple represents genes that did not
461 match any B73 gene model annotation.
462

463


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.15.766802; this version posted November 18, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.15.766802)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


464 **Literature cited**

465 Astashyn, A., Tvedte, E. S., Sweeney, D., Sapojnikov, V., Bouk, N., Joukov, V., Mozes, E.,
466 Strope, P. K., Sylla, P. M., Wagner, L., Bidwell, S. L., Brown, L. C., Clark, K., Davis, E.
467 W., Smith-White, B., Hlavina, W., Pruitt, K. D., Schneider, V. A., &amp; Murphy, T. D. (2024).
468 Rapid and sensitive detection of genome contamination at scale with FCS-GX. _Genome_
469 _Biology_, _25_ (1). https://doi.org/10.1186/s13059-024-03198-7
470 Brown, M. R., de La Rosa, P. M. G., &amp; Blaxter, M. (2025). tidk: a toolkit to rapidly identify
471 telomeric repeats from genomic datasets. _Bioinformatics_, _41_ (2).
472 https://doi.org/10.1093/bioinformatics/btaf049
473 Chen, Q., Washburn, J. D., Lima, D. C., Romay, M. C., Gage, J. L., Holland, J. B., Xavier, A.,
474 Murray, S. C., Ertl, D., Lopez-Cruz, M., de los Campos, G., Aguate, F. M., Beissinger, T.
475 M., Bohn, M. O., Buckler, E. S., Edwards, J., Flint-Garcia, S. A., Gore, M. A., Hirsch, C.
476 N., … de Leon, N. (2026). Genomes to fields 2024 maize genotype by environment
477 prediction competition. In _BMC Research Notes_ (Vol. 19, Number 1). BioMed Central Ltd.
478 https://doi.org/10.1186/s13104-026-07629-5
479 Cheng, H., Asri, M., Lucas, J., Koren, S., &amp; Li, H. (2024). Scalable telomere-to-telomere
480 assembly for diploid and polyploid genomes with double graph. _Nature Methods_, _21_ (6),
481 967–970. https://doi.org/10.1038/s41592-024-02269-8
482 Cheng, H., Concepcion, G. T., Feng, X., Zhang, H., &amp; Li, H. (2021). Haplotype-resolved de
483 novo assembly using phased assembly graphs with hifiasm. _Nature Methods_, _18_ (2), 170–
484 175. https://doi.org/10.1038/s41592-020-01056-5
485 Crow, T., Ta, J., Nojoomi, S., Aguilar-Rangel, M. R., Rodríguez, J. V. T., Gates, D., Rellán486 Álvarez, R., Sawers, R., &amp; Runcie, D. (2020). Gene regulatory effects of a large
487 chromosomal inversion in highland maize. _PLoS Genetics_, _16_ (12 December).
488 https://doi.org/10.1371/journal.pgen.1009213
489 Danecek, P., Bonfield, J. K., Liddle, J., Marshall, J., Ohan, V., Pollard, M. O., Whitwham, A.,
490 Keane, T., McCarthy, S. A., &amp; Davies, R. M. (2021). Twelve years of SAMtools and
491 BCFtools. _GigaScience_, _10_ (2). https://doi.org/10.1093/gigascience/giab008
492 Ellinghaus, D., Kurtz, S., &amp; Willhoeft, U. (2008). LTRharvest, an efficient and flexible software
493 for de novo detection of LTR retrotransposons. _BMC Bioinformatics_, _9_ (1), 18.
494 https://doi.org/10.1186/1471-2105-9-18
495 Flynn, J. M., Hubley, R., Goubert, C., Rosen, J., Clark, A. G., Feschotte, C., &amp; Smit, A. F.
496 (2020). RepeatModeler2 for automated genomic discovery of transposable element families.
497 _Proceedings of the National Academy of Sciences_, _117_ (17), 9451–9457.
498 https://doi.org/10.1073/pnas.1921046117
499 Foley, T. (2001). _Inbred corn line LH287. US Patent 6,281,414 B1_ .
500 https://patentimages.storage.googleapis.com/ae/88/78/be1f4ce2019a8f/US6281414.pdf
501 Fustier, M. A., Martínez-Ainsworth, N. E., Aguirre-Liguori, J. A., Venon, A., Corti, H.,
502 Rousselet, A., Dumas, F., Dittberner, H., Camarena, M. G., Grimanelli, D., Ovaskainen, O.,
503 Falque, M., Moreau, L., de Meaux, J., Montes-Hernández, S., Eguiarte, L. E., Vigouroux,
504 Y., Manicacci, D., &amp; Tenaillon, M. I. (2019). Common gardens in teosintes reveal the
505 establishment of a syndrome of adaptation to altitude. _PLoS Genetics_, _15_ (12).
506 https://doi.org/10.1371/journal.pgen.1008512
507 Ghaffari, R., Cannon, E. K. S., Kanizay, L. B., Lawrence, C. J., &amp; Dawe, R. K. (2013). Maize
508 chromosomal knobs are located in gene-dense areas and suppress local recombination.
509 _Chromosoma_, _122_ (1–2), 67–75. https://doi.org/10.1007/s00412-012-0391-8


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.15.766802; this version posted November 18, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.15.766802)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


510 Holst, F., Bolger, A. M., Kindel, F., Günther, C., Maß, J., Triesch, S., Kiel, N., Saadat, N.,
511 Ebenhöh, O., Usadel, B., Schwacke, R., Weber, A. P. M., Bolger, M. E., &amp; Denton, A. K.
512 (2026). Helixer: ab initio prediction of primary eukaryotic gene models combining deep
513 learning and a hidden Markov model. _Nature Methods_, _23_ (4), 732–739.
514 https://doi.org/10.1038/s41592-025-02939-1
515 Hufford, M. B., Seetharam, A. S., Woodhouse, M. R., Chougule, K. M., Ou, S., Liu, J., Ricci, W.
516 A., Guo, T., Olson, A., Qiu, Y., Coletta, R. Della, Tittes, S., Hudson, A. I., Marand, A. P.,
517 Wei, S., Lu, Z., Wang, B., Tello-Ruiz, M. K., Piri, R. D., … Dawe, R. K. (2021). De novo
518 assembly, annotation, and comparative analysis of 26 diverse maize genomes. _Science_,
519 _373_ (6555), 655–662. https://doi.org/10.1126/science.abg5289
520 Irber, L., Pierce-Ward, N. T., Abuelanin, M., Alexander, H., Anant, A., Barve, K., Baumler, C.,
521 Botvinnik, O., Brooks, P., Dsouza, D., Gautier, L., Hera, M. R., Houts, H. E., Johnson, L.
522 K., Klötzl, F., Koslicki, D., Lim, M., Lim, R., Nelson, B., … Brown, C. T. (2024).
523 sourmash v4: A multitool to quickly search, compare,and analyze genomic and
524 metagenomic data sets. _Journal of Open Source Software_, _9_ (98), 6830.
525 https://doi.org/10.21105/joss.06830
526 Kanizay, L. B., Albert, P. S., Birchler, J. A., &amp; Kelly Dawe, R. (2013). Intragenomic conflict
527 between the two major knob repeats of maize. _Genetics_, _194_ (1), 81–89.
528 https://doi.org/10.1534/genetics.112.148882
529 Kiełbasa, S. M., Wan, R., Sato, K., Horton, P., &amp; Frith, M. C. (2011). Adaptive seeds tame
530 genomic sequence comparison. _Genome Research_, _21_ (3), 487–493.
531 https://doi.org/10.1101/gr.113985.110
532 Kim, D., Paggi, J. M., Park, C., Bennett, C., &amp; Salzberg, S. L. (2019). Graph-based genome
533 alignment and genotyping with HISAT2 and HISAT-genotype. _Nature Biotechnology_,
534 _37_ (8), 907–915. https://doi.org/10.1038/s41587-019-0201-4
535 Li, H. (2018). Minimap2: Pairwise alignment for nucleotide sequences. _Bioinformatics_, _34_ (18),
536 3094–3100. https://doi.org/10.1093/bioinformatics/bty191
537 Lima, D. C., Washburn, J. D., Varela, J. I., Chen, Q., Gage, J. L., Romay, M. C., Holland, J.,
538 Ertl, D., Lopez-Cruz, M., Aguate, F. M., de los Campos, G., Kaeppler, S., Beissinger, T.,
539 Bohn, M., Buckler, E., Edwards, J., Flint-Garcia, S., Gore, M. A., Hirsch, C. N., … de
540 Leon, N. (2023). Genomes to Fields 2022 Maize genotype by Environment Prediction
541 Competition. _BMC Research Notes_, _16_ (1). https://doi.org/10.1186/s13104-023-06421-z
542 Liu, J., Seetharam, A. S., Chougule, K., Ou, S., Swentowsky, K. W., Gent, J. I., Llaca, V.,
543 Woodhouse, M. R., Manchanda, N., Presting, G. G., Kudrna, D. A., Alabady, M., Hirsch, C.
544 N., Fengler, K. A., Ware, D., Michael, T. P., Hufford, M. B., &amp; Dawe, R. K. (2020).
545 Gapless assembly of maize chromosomes using long-read technologies. _Genome Biology_,
546 _21_ (1). https://doi.org/10.1186/s13059-020-02029-9
547 Mölder, F., Jablonski, K. P., Letcher, B., Hall, M. B., Tomkins-Tinch, C. H., Sochat, V., Forster,
548 J., Lee, S., Twardziok, S. O., Kanitz, A., Wilm, A., Holtgrewe, M., Rahmann, S., Nahnsen,
549 S., &amp; Köster, J. (2021). Sustainable data analysis with Snakemake. _F1000Research_, _10_, 33.
550 https://doi.org/10.12688/f1000research.29032.1
551 Ou, S., Su, W., Liao, Y., Chougule, K., Agda, J. R. A., Hellinga, A. J., Lugo, C. S. B., Elliott, T.
552 A., Ware, D., Peterson, T., Jiang, N., Hirsch, C. N., &amp; Hufford, M. B. (2019).
553 Benchmarking transposable element annotation methods for creation of a streamlined,
554 comprehensive pipeline. _Genome Biology_, _20_ (1), 275. https://doi.org/10.1186/s13059-019555 1905-y


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.15.766802; this version posted November 18, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.15.766802)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


556 Pfund, J. (2002). _Inbred corn plant FBLL. US Patent 6,362,403 B1._
557 https://patentimages.storage.googleapis.com/b9/a0/41/af1347ffcb414d/US6362403.pdf
558 Quinlan, A. R., &amp; Hall, I. M. (2010). BEDTools: a flexible suite of utilities for comparing
559 genomic features. _Bioinformatics_, _26_ (6), 841–842.
560 https://doi.org/10.1093/bioinformatics/btq033
561 Rhie, A., Walenz, B. P., Koren, S., &amp; Phillippy, A. M. (2020). Merqury: Reference-free quality,
562 completeness, and phasing assessment for genome assemblies. _Genome Biology_, _21_ (1).
563 https://doi.org/10.1186/s13059-020-02134-9
564 Rogers, A. R., Bian, Y., Krakowsky, M., Peters, D., Turnbull, C., Nelson, P., &amp; Holland, J. B.
565 (2022). Genomic prediction for the Germplasm Enhancement of Maize project. _The Plant_
566 _Genome_, _15_ (4). https://doi.org/10.1002/tpg2.20267
567 Romero Navarro, J. A., Willcox, M., Burgueño, J., Romay, C., Swarts, K., Trachsel, S.,
568 Preciado, E., Terron, A., Delgado, H. V., Vidal, V., Ortega, A., Banda, A. E., Montiel, N.
569 O. G., Ortiz-Monasterio, I., Vicente, F. S., Espinoza, A. G., Atlin, G., Wenzl, P., Hearne, S.,
570 &amp; Buckler, E. S. (2017). A study of allelic diversity underlying flowering-time adaptation in
571 maize landraces. _Nature Genetics_, _49_ (3), 476–480. https://doi.org/10.1038/ng.3784
572 Schnable, P. S., Ware, D., Fulton, R. S., Stein, J. C., Wei, F., Pasternak, S., Liang, C., Zhang, J.,
573 Fulton, L., Graves, T. A., Minx, P., Reily, A. D., Courtney, L., Kruchowski, S. S.,
574 Tomlinson, C., Strong, C., Delehaunty, K., Fronick, C., Courtney, B., … Wilson, R. K.
575 (2009). The B73 Maize Genome: Complexity, Diversity, and Dynamics. _Science_,
576 _326_ (5956), 1112–1115. https://doi.org/10.1126/science.1178534
577 Shen, W., Le, S., Li, Y., &amp; Hu, F. (2016). SeqKit: A cross-platform and ultrafast toolkit for
578 FASTA/Q file manipulation. _PLoS ONE_, _11_ (10).
579 https://doi.org/10.1371/journal.pone.0163962
580 Sommer, M. J., Zimin, A. V., &amp; Salzberg, S. L. (2025). PSAURON: a tool for assessing protein
581 annotation across a broad range of species. _NAR Genomics and Bioinformatics_, _7_ (1).
582 https://doi.org/10.1093/nargab/lqae189
583 Song, B., Marco-Sola, S., Moreto, M., Johnson, L., Buckler, E. S., &amp; Stitzer, M. C. (2022).
584 AnchorWave: Sensitive alignment of genomes with high sequence diversity, extensive
585 structural polymorphism, and whole-genome duplication. _Proceedings of the National_
586 _Academy of Sciences_, _119_ (1). https://doi.org/10.1073/pnas.2113075119
587 Su, W., Gu, X., &amp; Peterson, T. (2019). TIR-Learner, a New Ensemble Method for TIR
588 Transposable Element Annotation, Provides Evidence for Abundant New Transposable
589 Elements in the Maize Genome. _Molecular Plant_, _12_ (3), 447–460.
590 https://doi.org/10.1016/j.molp.2019.02.008
591 Tan, K., Liu, X., Wang, Z., Zhang, Z., Huang, W., Liu, S., Lin, Z., Zhao, H., Zhao, H., Liu, Y.,
592 Han, F., Lai, J., Song, W., Zhao, J., &amp; Chen, J. (2026). Near‐complete genome assembly of
593 a transformation‐efficient elite inbred line LH244 and its comparison with B73. _Journal of_
594 _Integrative Plant Biology_, _68_ (2), 366–382. https://doi.org/10.1111/jipb.70099
595 Tang, H., Krishnakumar, V., Zeng, X., Xu, Z., Taranto, A., Lomas, J. S., Zhang, Y., Huang, Y.,
596 Wang, Y., Yim, W. C., Zhang, J., &amp; Zhang, X. (2024). JCVI: A versatile toolkit for
597 comparative genomics analysis. _iMeta_, _3_ (4). https://doi.org/10.1002/imt2.211
598 Tang, H., Zhang, X., Miao, C., Zhang, J., Ming, R., Schnable, J. C., Schnable, P. S., Lyons, E., &amp;
599 Lu, J. (2015). ALLMAPS: Robust scaffold ordering based on multiple maps. _Genome_
600 _Biology_, _16_ (1). https://doi.org/10.1186/s13059-014-0573-1


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.15.766802; this version posted November 18, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.15.766802)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


601 Tegenfeldt, F., Kuznetsov, D., Manni, M., Berkeley, M., Zdobnov, E. M., &amp; Kriventseva, E. V.
602 (2025). OrthoDB and BUSCO update: annotation of orthologs with wider sampling of
603 genomes. _Nucleic Acids Research_, _53_ (D1), D516–D522.
604 https://doi.org/10.1093/nar/gkae987
605 National Plant Germplasm System. (2025). _National Plant Germplasm System_ .
606 https://npgsweb.ars-grin.gov/gringlobal/search
607 USDA Plant Variety Protection Office. (2000). Plant Variety Protection Certificate No. 9700218:
608 PH09B (Zea mays L.). In _U.S. Department of Agriculture, Agricultural Marketing Service_ .
609 U.S. Department of Agriculture, Agricultural Marketing Service.
610 USDA Plant Variety Protection Office. (2001). _Plant Variety Protection Certificate No._
611 _9900382: PH2MW (Zea mays L.)_ . U.S. Department of Agriculture, Agricultural Marketing
612 Service. http://www.ams.usda.gov/science/pvp.htm
613 White, M. R., Mikel, M. A., de Leon, N., &amp; Kaeppler, S. M. (2020). Diversity and heterotic
614 patterns in North American proprietary dent maize germplasm. _Crop Science_, _60_ (1), 100–
615 114. https://doi.org/10.1002/csc2.20050
616 Xiong, W., He, L., Lai, J., Dooner, H. K., &amp; Du, C. (2014). HelitronScanner uncovers a large
617 overlooked cache of _Helitron_ transposons in many plant genomes. _Proceedings of the_
618 _National Academy of Sciences_, _111_ (28), 10263–10268.
619 https://doi.org/10.1073/pnas.1410068111
620 Xu, Z., &amp; Wang, H. (2007). LTR_FINDER: an efficient tool for the prediction of full-length
621 LTR retrotransposons. _Nucleic Acids Research_, _35_ (Web Server), W265–W268.
622 https://doi.org/10.1093/nar/gkm286
623 Yao, W., Li, S., Ren, J., Shi, J., Wu, F., Zhao, S., Su, H., Wang, Y., Chen, Y., Ren, Z., &amp; Ku, L.
624 (2026). Chromosome-level genome assemblies of two maize inbred lines with contrasting
625 plant architectures. _Scientific Data_, _13_ (1), 276. https://doi.org/10.1038/s41597-026-06603-x
626

627 **Data availability**
628 Raw sequencing reads are available through the European Nucleotide Archive (ENA) under
629 accession PRJEB82638. Genome assemblies are available through MaizeGDB under the
630 following assembly identifiers:
631 - **PHBW8:** Zm-PHWB8-DRAFT-G2F-1.0

632 - **FBLL:** Zm-FBLL-DRAFT-G2F-1.0

633 - **3AZA1:** Zm-3AZA1-DRAFT-G2F-1.0

634 - **PHB47:** Zm-PHB47-DRAFT-G2F-1.0

635 - **PHP02:** Zm-PHP02-DRAFT-G2F-1.0

636 - **LH287:** Zm-LH287-DRAFT-G2F-1.0

637 - **GEMS-0113:** Zm-GEMS_0113-DRAFT-G2F-1.0
638 - **GEMS-0061:** Zm-GEMS_0061-DRAFT-G2F-1.0
639 - **GEMS-0219:** Zm-GEMS_0219-DRAFT-G2F-1.0
640 - **GEMS-0227:** Zm-GEMS_0227-DRAFT-G2F-1.0
641 **Code availability**
642 All computational methods for assembly are documented and scripted for public use on GitHub
643 &#91;(https://github.com/maize-genetics/g2f_2024).&#93;(https://github.com/maize-genetics/g2f_2024)
644 **Acknowledgements**
645 We acknowledge the Cornell BioHPC (RRID:SCR_021757) for computational resources used in
646 this work.


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.15.766802; this version posted November 18, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.15.766802)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


647 **Funding**
648 This work was supported in part by Circular Economy that Reimagines Corn Agriculture
649 (CERCA) Nitrogen grant: Foundation for Food &amp; Agriculture Research (FFAR) with additional
650 support from the Grantham Foundation.
651 **Author contributions**
652 C.E.G., E.S.B., M.H., and M.C.R. conceived the project. N.A.J., C.E.G., and T.P.M. assembled
653 genomes. A.M. annotated genomes. N.A.J. and A.M. conducted comparative genomics analyses.
654 All authors contributed to manuscript drafting and revisions.
655 **Conflicts of interest**

656 The authors declare no conflicts of interest.

657


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.15.766802; this version posted November 18, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.15.766802)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


658 **Supplementary data**
659 **Supplementary Table 1: PacBio HiFi and Oxford Nanopore sequencing inputs for thirteen**
660 **maize lines.**

**Genotype** **HiFi (Gb) HiFi coverage (×) ONT ≥50 kb (Gb) ONT coverage (×)**

3AZA1 150.05 65.2 0.216 0.09

FBLL 97.03 42.2 0.974 0.42

GEMS-0061 79.36 34.5 0.187 0.08

GEMS-0113 74.19 32.3 0.049 0.02

GEMS-0219 57.05ᵃ 24.8 –ᵇ –ᵇ

GEMS-0227 103.51 45.0 0.048 0.02

LH244 NA NA NA NA

LH287 115.74ᶜ 50.3 –ᶜ –ᶜ

PH09B 82.76ᶜ 36.0 –ᶜ –ᶜ

PH2MW 69.31ᶜ 30.1 –ᶜ –ᶜ

PHB47 69.68 30.3 0.032 0.01

PHBW8 84.31 36.7 0.742 0.32

PHP02 75.87 33.0 1.952 0.85

661 _Sequencing amounts represent reads used as assembly inputs, as verified from assembly_
662 _provenance or reported in collaborator documentation. ONT inputs consist of reads ≥50 kb._
663 _Coverage assumes a genome size of 2.3 Gb for both technologies. Inputs for partially phased_
664 _accessions GEMS-0061 and GEMS-0113 are reported once per accession. NA indicates_
665 _unavailable sequencing records; – indicates no ONT assembly input documented._
666 _ᵃ HiFi reads from a single SMRT cell; two additional cells, derived from a different plant of the_
667 _same accession, were excluded._
668 _ᵇ ONT reads were generated but excluded from assembly because their plant of origin relative to_
669 _the HiFi cell used could not be verified._
670 _ᶜ HiFi quantities are from collaborator-provided read sets. Provider documentation describes_
671 _HiFi-only assemblies, although ONT reads were also generated. ONT usage remains to be_
672 _independently confirmed._
673

674


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.15.766802; this version posted November 18, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.15.766802)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


675 **Supplementary Table 2:** **Summary** **statistics for the** **genomic features identified**















676

677


|Genotype|MeanCDS&lt;br&gt;length(bp)|MeanProtein&lt;br&gt;length(aa)|MeanGene&lt;br&gt;length(bp)|Mean&lt;br&gt;Exonsper&lt;br&gt;gene|
|---|---|---|---|---|
|FBLL|1136.5|377.8|4276.9|5.88|
|PHBW8|1145.2|380.7|4290.7|5.9|
|PHB47|1138.5|378.5|4291.9|5.88|
|3AZA1|1136.2|377.7|4278.8|5.89|
|PHP02|1138.8|378.6|4294.6|5.9|
|GEMS-0227|1137.2|378.1|4271.4|5.89|
|GEMS-0219|1138.1|378.4|4272.5|5.89|
|GEMS-0061_hap1|1134.8|377.3|4253.1|5.86|
|GEMS-0061_hap2|1134.4|377.1|4265.8|5.87|
|GEMS-0113_hap1|1140.9|379.3|4286|5.9|
|GEMS-0113_hap2|1141.6|379.5|4302.8|5.91|
|LH287|1140.6|379.2|4293.8|5.88|
|PH09B|1138.5|378.5|4263.7|5.87|
|PH2MW|1139|378.7|4273.7|5.87|
|LH244|1137.7|378.2|4268.7|5.88|


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.15.766802; this version posted November 18, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.15.766802)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


678 **Supplementary Table 3. BUSCO completeness of the Helixer gene annotations**
679
























|Genotype|Complete&lt;br&gt;(C %)&lt;br&gt;,|Single-copy&lt;br&gt;(S %)&lt;br&gt;,|Duplicated&lt;br&gt;(D %)&lt;br&gt;,|Fragmented&lt;br&gt;(F %)&lt;br&gt;,|Missing&lt;br&gt;(M %)&lt;br&gt;,|
|---|---|---|---|---|---|
|FBLL|97.3|72.2|25.1|1.3|1.4|
|PHBW8|97.1|71.7|25.4|1.3|1.5|
|PHB47|97.2|72.3|24.9|1.3|1.5|
|3AZA1|97|72|25|1.4|1.6|
|PHP02|97.1|71.7|25.4|1.4|1.4|
|GEMS-0227|97.1|70.1|27|1.3|1.5|
|GEMS-0219|97|71.1|25.9|1.2|1.8|
|GEMS-0061 hap1|96.6|70.4|26.2|1.3|2.1|
|GEMS-0061 hap2|96.2|70.1|26.1|1.4|2.4|
|GEMS-0113 hap1|96.3|70.5|25.8|1.3|2.4|
|GEMS-0113 hap2|95.9|71.1|24.8|1.2|2.8|
|LH287|96.9|71.9|25|1.5|1.6|
|PH09B|97.2|71.7|25.5|1.2|1.6|
|PH2MW|97|71.8|25.3|1.3|1.6|
|LH244|97|72.1|24.9|1.4|1.6|


680 **Supplementary Table 4: Summary statistics of PSAURON assessment of predicted gene models across the genotypes**


681






































|Col1|Col2|Col3|Col4|Genemodelswithscore&lt;05(%)&lt;br&gt;.|Col6|Col7|Col8|Col9|
|---|---|---|---|---|---|---|---|---|
|**Genotype**|**Genes**|**Protein-**&lt;br&gt;**coding (%)**|**Mean in-**&lt;br&gt;**frame score**|**CDS**&lt;br&gt;**&lt;300 bp**|**CDS**&lt;br&gt;**300–900**&lt;br&gt;**bp**|**CDS**&lt;br&gt;**≥900 bp**|**Single-**&lt;br&gt;**exon**|~~682~~&lt;br&gt;683&lt;br&gt;**Multi-**&lt;br&gt;**exon**|
|FBLL|40677|93.9|0.908|21.3|8.9|0.5|10.3|~~684~~&lt;br&gt;5.3|
|PHBW8|40563|93.9|0.909|20.8|9.3|0.6|10.1|&lt;br&gt;5.4|
|PHB47|40465|93.9|0.908|20.1|9.3|0.6|9.9|~~685~~&lt;br&gt;5.4|
|3AZA1|40594|93.7|0.907|22.1|9.3|0.6|10.4|&lt;br&gt;5.5|
|PHP02|40705|93.9|0.908|21.6|9|0.6|10.2|686&lt;br&gt;5.4|
|GEMS-0227|40723|93.9|0.909|21.7|9.1|0.5|10.4|5.3|
|GEMS-0219|40777|93.7|0.907|22|9.4|0.6|10.3|687&lt;br&gt;5.6|
|GEMS-0061 hap1|40873|93.6|0.906|21.8|9.5|0.6|10.5|&lt;br&gt;5.6|
|GEMS-0061 hap2|40724|93.7|0.906|21.4|9.4|0.5|10.5|~~688~~&lt;br&gt;5.5|
|GEMS-0113 hap1|40741|93.7|0.907|21|9.5|0.6|10.3|5.5|
|GEMS-0113 hap2|39868|93.8|0.908|21.2|9.5|0.5|10|5.5|
|LH287|39984|93.7|0.907|21.9|9.5|0.6|10.8|5.5|
|PH09B|40956|93.9|0.908|21.1|9.2|0.6|10.1|5.4|
|PH2MW|41074|93.8|0.908|21.6|9.2|0.7|10.2|5.5|
|LH244|40585|93.8|0.908|21.2|9.4|0.5|10.1|5.5|


689 **Supplementary Table 5: Summary of gene model comparisons to B73 across the genotypes using Liftoff**





















690


691


|Genotype|Genes&lt;br&gt;per&lt;br&gt;Genotype|Genes&lt;br&gt;matches|%gene&lt;br&gt;matched|Genotype&lt;br&gt;genes&lt;br&gt;matched|Genes&lt;br&gt;without&lt;br&gt;B73&lt;br&gt;match|B73&lt;br&gt;genes&lt;br&gt;withone&lt;br&gt;match|B73genes&lt;br&gt;with≥2&lt;br&gt;matches|Genes&lt;br&gt;matched&lt;br&gt;to≥2B73&lt;br&gt;genes|
|---|---|---|---|---|---|---|---|---|
|FBLL|40677|31155|79.8|31947|8730|30069|1086|464|
|PHBW8|40563|30439|78|31205|9358|29407|1032|427|
|PHB47|40465|30407|77.9|31188|9277|29357|1050|422|
|3AZA1|40594|30534|78.2|31261|9333|29532|1002|431|
|PHP02|40705|30337|77.7|31061|9644|29312|1025|430|
|GEMS-0227|40723|29990|76.8|30773|9950|28946|1044|421|
|GEMS-0219|40777|30637|78.5|31385|9392|29560|1077|471|
|GEMS-0061_hap1|40873|30426|77.9|31213|9660|29368|1058|431|
|GEMS-0061_hap2|40724|30265|77.5|31065|9659|29217|1048|416|
|GEMS-0113_hap1|40741|30482|78.1|31225|9516|29428|1054|463|
|GEMS-0113_hap2|39868|30251|77.5|30974|8894|29228|1023|448|
|LH287|39984|29792|76.3|30493|9491|28811|981|403|
|PH09B|40956|30729|78.7|31441|9515|29704|1025|460|
|PH2MW|41074|30408|77.9|31128|9946|29391|1017|425|
|LH244|40585|30977|79.4|31764|8821|29883|1094|479|


692 **Supplementary Tables 6 &amp; 7 in attached document.**


693


694 **Supplementary Table 8: Summary statistics for all the TE features identified**





























695


|Genotype|All&lt;br&gt;TEs&lt;br&gt;(%)|LTR:&lt;br&gt;intact&lt;br&gt;elements|LTR:&lt;br&gt;families|LTR:&lt;br&gt;content&lt;br&gt;(%)|TIR:&lt;br&gt;intact&lt;br&gt;elements|TIR:&lt;br&gt;families|Helitron:&lt;br&gt;intact&lt;br&gt;elements|Helitron:&lt;br&gt;families|LINE:&lt;br&gt;families|SINE:&lt;br&gt;families|
|---|---|---|---|---|---|---|---|---|---|---|
|FBLL|88.28|58,727|10,360|78.98|21,892|4,972|4,357|2,312|17|0|
|PHBW8|88.23|58,576|10,684|79.09|21,732|5,227|4,393|2,303|15|0|
|PHB47|87.02|58,254|10,513|78.46|21,435|5,120|4,417|1,528|16|0|
|3AZA1|88.61|59,741|10,119|79.59|21,832|5,359|4,398|2,356|17|0|
|PHP02|87.89|58,665|9,840|78.79|20,516|4,844|4,455|1,932|19|0|
|GEMS-0227|87.53|59,635|10,219|78.53|21,767|5,149|4,366|1,991|14|0|
|GEMS-0219|87.78|59,143|10,667|78.29|21,549|5,049|4,404|2,238|10|0|
|GEMS-0061&lt;br&gt;hap1|86.93|59,392|8,752|78.13|21,093|5,218|4,470|1,931|19|0|
|GEMS-0061&lt;br&gt;hap2|88.2|58,394|10,729|79.31|21,269|5,180|4,451|2,175|18|0|
|GEMS-0113&lt;br&gt;hap1|88.08|59,045|10,501|78.76|21,786|4,863|4,408|2,148|21|0|
|GEMS-0113&lt;br&gt;hap2|87.72|58,730|10,253|78.74|21,653|4,934|4,341|2,370|17|0|
|LH287|83.63*|54,183|9,529|76.89|20,982|4,612|NA|NA|12|0|
|PH09B|86.30*|59,545|10,236|79.17|21,735|4,704|NA|NA|17|0|
|PH2MW|86.02*|58,561|10,418|79.01|20,847|4,990|NA|NA|15|0|
|LH244|87.44|59,439|10,695|78.79|21,879|5,023|4,418|1,291|13|0|


