Drosophila melanogaster pangenome (minigraph on dm6), indexed for JBrowse 2 ========================================================================== A graph built here from 29 published NCBI assemblies and the dm6 reference, plus tabix-indexed BED projections of it, so that JBrowse can query a locus without downloading the graph. No graph was published for this panel; this one was constructed. See "How it was built". Assemblies ---------- 30 sequence sets, all fetched from NCBI with the datasets CLI (datasets 18.22.1) on 2026-11-13: dm6 as the reference backbone and 29 chromosome-level D. melanogaster assemblies, the panel a survey of NCBI found at contig N50 18 Mb or more. name accession source dm6 GCF_000001215.4 ISO1, BDGP Release 6 + ISO1 MT (RefSeq) KSA2 GCA_003401745.1 } Chakraborty et al. 2019, Nat Commun QI2 GCA_003402055.1 } 10:4872, doi:10.1038/s41467-019-12884-1, Canton-S_A1 GCA_003401735.1 } PRJNA418342: 13 Drosophila Synthetic BER1_B1 GCA_003401925.1 } Population Resource founders and BS1_A3 GCA_003401795.1 } Oregon-R, PacBio CLR. The paper's Wild5B_A6 GCA_003401885.1 } founder codes: KSA2 is A4, QI2 B3, A7 GCA_003401915.1 } OreR its Ore; the code follows the SAMry506_AB8 GCA_003401685.1 } other names OreR GCA_003402015.1 } BOG1_A2 GCA_003401805.1 } VAG1_A5 GCA_003401855.1 } RVC3_B4 GCA_003402005.1 } CA1_B2 GCA_003401975.1 } B6 GCA_003397115.2 } BDSC_5295 GCA_055682045.1 } Samano et al. 2026, Genome Res, BDSC_620 GCA_055681985.1 } doi:10.1101/gr.281434.125, BDSC_1349 GCA_056149345.1 } PRJNA1214913: 11 Bloomington stocks BDSC_1282 GCA_055723745.1 } carrying classical visible markers, BDSC_554 GCA_055768685.1 } Oxford Nanopore, named by Bloomington BDSC_6027 GCA_056149365.1 } stock number BDSC_2969 GCA_055768665.1 } BDSC_156 GCA_055723785.1 } BDSC_1570 GCA_055723755.1 } BDSC_576 GCA_055768645.1 } BDSC_662 GCA_055768715.1 } A4_HiFi GCA_059582535.1 } PRJNA1482540, single-fly PacBio HiFi GUM_HiFi GCA_059582545.1 } (Texas A&M, 2026); whether A4 and SAM SAM_HiFi GCA_059582525.1 } are the DSPR founders of those names } is not established CantonS_Yunnan GCA_048772135.1 Canton S, ONT and PacBio, PRJNA1237537 That order, the reference first, is the order the genomes were folded into the graph. Each assembly's sequence_report.jsonl says which of its sequences is which chromosome; every sequence whose role is assembled-molecule was kept and renamed #1#chr (PanSN; MT became chrM, UCSC's spelling), and unplaced and unlocalized scaffolds were left out. sources.sha256.tsv beside this file holds the SHA-256 of every genomic.fna as NCBI served it. What each assembly contributes: dm6 chr2L chr2R chr3L chr3R chr4 chrX chrY chrM the 14 Chakraborty chr2L chr2R chr3L chr3R chr4 chrX chrM the 11 BDSC chr2L chr2R chr3L chr3R chr4 chrX A4/GUM/SAM_HiFi chr2L chr2R chr3 chr4 chrX CantonS_Yunnan chr2L chr2R chr3L chr3R chr4 chrX chrY The three HiFi assemblies name chromosome 3 as one molecule ("3", 64-66 Mb) where dm6 has 3L and 3R; it is kept as chr3, and minigraph aligns it to both arms. The BDSC and HiFi assemblies' chrX runs 26-38 Mb against dm6's 23.5 Mb, since they carry more of the heterochromatin. Reference --------- UCSC dm6 == NCBI GCF_000001215.4 (Release 6 plus ISO1 MT), whose RefSeq chromosomes were renamed to UCSC's names through the sequence report. The build refuses unless each of the eight has dm6.chrom.sizes' length. https://jbrowse.org/ucsc/dm6/config.json How it was built ---------------- minigraph 0.21-r606, whole genome, dm6 first and then the panel in the order above: minigraph -cxggs -t 8 dm6.pansn.fa KSA2.pansn.fa ... CantonS_Yunnan.pansn.fa It took 69,389 s (19.3 hours) on 8 threads. The graph has 129,369 segments and 185,455 links, 264,316,580 bp in all; its rank-0 segments, dm6's eight chromosomes, sum to 137,567,484 bp. The build stops unless every SN tag is #1#chr<2L|2R|3L|3R|3|4|X|Y|M>, every segment id is unique, and rank 0 is dm6 alone, each chromosome at its dm6.chrom.sizes length. Files ----- fly-dm6-minigraph.rgfa.gz the graph itself (bgzipped rGFA) fly-dm6-minigraph.graph.bed.gz (+ .tbi) nodes and links in one file, gfa-to-tabix's format 2 (built with gfa-to-tabix 4.0.0). RgfaTabixAdapter takes the prefix `fly-dm6-minigraph` with "format": 2 fly-dm6-minigraph.fold10000.graph.bed.gz coarse tier: the graph with (+ .tbi) variants under 10 kb folded into the reference, so a whole chromosome draws fly-dm6-minigraph.bubbles.bed.gz gfatools bubble output, indexed (+ .csi, .tbi) both ways: the pangenome configs read the .tbi sources.sha256.tsv name, accession and SHA-256 of each assembly's genomic.fna checksums.sha256 SHA-256 of each file above build.log the run's log Caveats ------- - The graph holds no path per assembly. minigraph -cxggs writes no P or W lines, so a segment's rank (SR) is the order in which minigraph added it, never which strains carry it. The structural-state sidecar the genomes.jbrowse.org pangenome page reads (https://jbrowse.org/pangenome/fly-dm6/sv-states/) recovers carriage per bubble by mapping each assembly back with minigraph --call; its own README says how. - The panel is three sets with their own purposes, not a population sample: inbred lab founders from a worldwide sample, marker stocks crossed to combine visible mutations, and single flies. A share of the panel is not a population frequency. - chrY holds dm6 and CantonS_Yunnan, chrM dm6 and the 14 Chakraborty strains, so a lane there is missing for the rest because they lack the sequence, not because it was dropped. - Heterochromatin and the unplaced dm6 scaffolds are outside the graph. Reproduce --------- bash scripts/build_fly_pangenome.sh (GMOD/jbrowse-components) It needs the NCBI datasets CLI, minigraph, gfatools, gfa-to-tabix, htslib and python3, and skips each stage whose output exists. Licence and attribution ----------------------- The assemblies are public NCBI GenBank and RefSeq records; cite their papers and accessions, not this file. The graph and the projections here are derived work produced for the JBrowse 2 demos and carry no additional restriction.