1001 Genomes Plus Arabidopsis accessions vs TAIR10: SyRI lanes and a minigraph pangenome, indexed for JBrowse 2 ============================================================================== Two views of the same 27 genomes, built here from the 1001 Genomes Plus Phase 1 release: SyRI's classification of every accession against TAIR10 (syntenic, inverted, translocated, duplicated regions), loaded as multi-genome lanes, and a minigraph pangenome of the 27 with the tabix projections JBrowse queries a locus from. The accessions' own gene, transposon, methylation and histone tracks are not rehosted: config.json points at the 1001 Genomes data centre, which serves them with CORS and Range support. Its .{CG,CHG,CHH}meth.bw files store each cytosine as a zero-length record, so their zoom levels hold no sums and the methylation tracks read zero from 80 bp/px; zoomed in they draw. Assemblies ---------- TAIR10 (GCF_000001735.4, NCBI RefSeq), the five nuclear chromosomes as Chr1 to Chr5, plus the 26 PacBio CLR assemblies of 1001G+ Phase 1 (Igolkina et al., Nature Genetics 2025, doi:10.1038/s41588-025-02293-0), fetched from https://1001genomes.org/data/1001Gp/27genomes/releases/current/assemblies/ .scaffolds_corrected.v2.1.fasta.gz on 2026-09-24. Each holds its five chromosomes as _Chr1 to _Chr5 and nothing else; they were renamed Chr1 to Chr5 for SyRI, which pairs chromosomes by name, and #1#Chr1 to Chr5 (PanSN) for the graph. The .aliases.txt beside each chrom.sizes maps the portal's names back, so the portal's own tracks load on these assemblies unchanged. id name country admixture group (1001 Genomes accession table) 6909 Col-0.6909 USA germany the reference's own accession 1741 KBS-Mac-74 USA germany 6966 Sq-1 UK western_europe 10002 TueWa1-2 GER western_europe 8236 HSm CZE central_europe 9728 Stiav-1 SVK central_europe 6024 Fly2-2 SWE south_sweden 6124 T690 SWE south_sweden 6069 Nyl-7 SWE north_sweden 6244 TRA-01 SWE north_sweden 9981 Angit-1 ITA italy_balkan_caucasus 9075 Lerik1-4 AZE italy_balkan_caucasus 9537 IP-Cum-1 ESP spain 9888 IP-Pva-1 ESP spain 9543 IP-Gra-0 ESP relict 9905 IP-Ven-0 ESP relict 9764 Qar-8a LBN admixed 9638 Noveg-3 RUS asia 22002 35-1 Morocco } outside the 1135 panel; Durvasula et al. 22001 85-3 Morocco } 2017 (doi:10.1073/pnas.1616736114) 22006 Areeiro-1 Madeira } 22007 ET-86.4 Ethiopia } 22004 Elh-2 Morocco } 22005 Rabacal-1 Madeira } 22003 Taz-0 Morocco } 10024 Tnz-1 Tanzania } That order, Col-0's own accession first and then by admixture group, is the lane order and the order the genomes were folded into the graph. How the SyRI lanes were built ----------------------------- Each accession against TAIR10: minimap2 -cx asm5 --eqx -t 5 TAIR10.fa .fa > TAIR10_.aln.paf syri -c TAIR10_.aln.paf -r TAIR10.fa -q .fa -F P --nc 5 minimap2 2.24-r1122; SyRI 1.8.2 installed from github.com/schneebergerlab/syri at acf72ccf47beffaaeade849fd258827ef24bedf4. scripts/syri_to_paf.py then wrote each pair's top-level regions (SYN, INV, TRANS, INVTR, DUP, INVDP) as PAF records between PanSN names carrying syri:Z: and color:Z:, concatenated into syri_1001g.paf, and the same regions on TAIR10 alone with the accession in a `query` column, concatenated into syri_regions.bed.gz. What SyRI found, per accession, against TAIR10: name syntenic inversions (bp) transloc. dup(ref) unaligned ref bp Col-0.6909 13 5 (34,741) 2 2 591,711 KBS-Mac-74 71 19 (477,888) 27 17 6,696,266 Sq-1 72 16 (1,644,903) 49 30 7,155,979 TueWa1-2 77 12 (1,921,791) 61 45 8,722,368 HSm 72 26 (2,094,818) 33 28 7,366,139 Stiav-1 37 13 (1,651,092) 13 2 4,042,977 Fly2-2 71 18 (1,599,358) 25 43 6,773,606 T690 83 12 (1,769,652) 42 25 6,216,695 Nyl-7 78 19 (3,797,059) 45 48 7,273,656 TRA-01 85 21 (1,960,790) 55 42 8,353,764 Angit-1 94 18 (1,709,288) 66 32 7,363,470 Lerik1-4 66 18 (1,713,860) 29 33 6,092,120 IP-Cum-1 78 18 (2,265,338) 47 27 6,213,962 IP-Pva-1 85 20 (1,894,512) 46 25 8,160,511 IP-Gra-0 97 22 (2,985,086) 46 32 7,757,951 IP-Ven-0 129 28 (3,546,861) 97 64 8,825,150 Qar-8a 96 21 (2,775,739) 48 38 8,381,861 Noveg-3 86 20 (2,151,463) 54 43 7,315,422 35-1 87 21 (1,248,642) 40 23 6,885,316 85-3 89 16 (1,678,830) 51 21 6,172,457 Areeiro-1 92 27 (4,515,752) 40 33 7,334,345 ET-86.4 116 28 (3,457,695) 48 47 7,382,713 Elh-2 106 21 (1,973,652) 87 40 8,697,636 Rabacal-1 114 29 (3,567,052) 67 40 8,101,758 Taz-0 103 24 (2,165,653) 40 54 8,413,812 Tnz-1 111 27 (2,785,232) 55 39 7,664,125 Col-0's own accession is the control: 13 syntenic regions cover the genome and its 5 inversions total 35 kb, where every other accession carries 12 to 29 inversions over 0.5 to 4.5 Mb. The chromosome 4 knob inversion (Fransz et al. 2000; Zapata et al. 2016): 24 of the 26 accessions carry an inverted region against TAIR10 inside Chr4:1.6-2.8 Mb, and in nine of them SyRI's interval is exactly Chr4:1,612,605-2,782,621 (1,170,016 bp), the same interval it reports between TAIR10 and Landsberg erecta. The two accessions with TAIR10's arrangement are Col-0's own assembly and KBS-Mac-74 (Michigan, USA). The reference carries the rare arrangement in this panel. How the graph was built ----------------------- minigraph 0.21-r606, TAIR10 first, then the accessions in lane order: minigraph -cxggs -t 16 TAIR10.pansn.fa Col-0.6909.pansn.fa ... Tnz-1.pansn.fa minigraph emits rGFA natively, so SN/SO/SR tags are read from the graph. Two audits ran on the result: every S line's SN tag matches #1#Chr[1-5], and the rank-0 thread is TAIR10's. gfatools stat (gfatools 0.5-r296): Number of segments: 192327 Number of links: 270208 Number of arcs: 540416 Max rank: 26 Total segment length: 211789247 Average segment length: 1101.194 Sum of rank-0 segment lengths: 119146348 Max degree: 7 Average degree: 1.405 The knob inversion is not in the graph as a bubble: the largest bubble inside Chr4:1.6-2.8 Mb spans 72 kb of TAIR10 (Chr4:1,778,767-1,850,855, longest allele 99 kb). SyRI's rows are where the knob shows; the graph holds the smaller variation inside and around it. The largest inversion bubble in the graph is Chr4:4,264,923-4,794,539 (530 kb). Projections, each tabix-indexed: scripts/build_rgfa_tabix.sh .segs.bed.gz, .links.bed.gz, and the .ref.* pair keyed under TAIR10 only scripts/build_rgfa_alleles.sh .alleles.bed.gz, one row per allele the graph holds, with a CIGAR for its size gfatools bubble .bubbles.bed.gz scripts/build_bubble_tier.sh .tier10000.segs.bed.gz, one node per bubble scripts/build_minigraph_paths.sh .paths.bed.gz, each accession's path through every bubble (minigraph --call per accession), rows by `strain` Other tracks on TAIR10 ---------------------- TAIR10.genes.gff.gz NCBI RefSeq annotation of GCF_000001735.4, renamed to Chr1 to Chr5, sorted, bgzipped fst.bw, omega.bw the 1001 Genomes project's 10 kb-window scans over the 1135 accessions: maximum Fst between admixture groups, and OmegaPlus sweep scores, from https://1001genomes.org/data/GMI-MPI/releases/current/fst_scans/ (the last window of each chromosome clipped to the chromosome end) RepeatMasker UCSC GenArk's bigBed for GCF_000001735.4, loaded over the NC_ aliases, not rehosted Reproduce --------- bash scripts/build_arabidopsis_pangenome.sh (GMOD/jbrowse-components). Everything above is what it writes. Terms ----- The 1001 Genomes Plus data are released by the 1001 Genomes consortium (https://1001genomes.org, contact via the site); cite Igolkina et al. 2025 and the accession table, not this redistribution. TAIR10 and its RefSeq annotation are public domain NCBI data.