refbook

One catalog for the many published versions of the human reference genome.

"GRCh38" names dozens of different FASTA files, and "hg19" or "b37" as many again: UCSC, Ensembl, GENCODE, the GRC's analysis sets, Broad/GATK, 1000 Genomes, iGenomes. They differ in which sequences they contain, how they name and order them, and whether some bases are masked or rewritten. refbook describes every one of them as a recipe applied to the same roots, the original INSDC sequences, and counts a version as cataloged only when its recipe reproduces the published file sequence by sequence.

Release 0.2.0 covers 55 published files, which reduce to 44 editions: distinct sets of named sequences, whatever their order. Each edition has a refbook label; the publishers' names for its files are its aliases.

The catalog is organized in 3 books, one per assembly: every release of the assembly, with the sequences its editions add from outside it, and the editions made from them. One set of roots serves every book: a sequence GRCh37 and GRCh38 both carry (rCRS, some unplaced scaffolds, GRCh37 patches that became GRCh38 alt loci) is one root.

Book	Assembly	Releases	Sequences in them	Editions	Published files
grch37	GRCh37	GRCh37 – GRCh37.p13 (14)	335	17	20
grch38	GRCh38	GRCh38 – GRCh38.p14 (15)	711	13	17
t2t-chm13	T2T-CHM13	T2T-CHM13v1.0 – T2T-CHM13v2.0 (3)	48	14	18

Edition	Assembly	Published as
b37	GRCh37	1000 Genomes human_g1k_v37
b37-ebv	GRCh37	Broad Institute Homo_sapiens_assembly19
ensembl-grch37-alt	GRCh37	Ensembl GRCh37 116 Homo_sapiens.GRCh37.dna.alt
ensembl-grch37-chromosomes	GRCh37	Illumina iGenomes Ensembl/GRCh37
ensembl-grch37-primary	GRCh37	Ensembl 75 Homo_sapiens.GRCh37.75.dna.primary_assembly; Ensembl GRCh37 116 Homo_sapiens.GRCh37.dna.primary_assembly
ensembl-grch37-toplevel	GRCh37	Ensembl 75 Homo_sapiens.GRCh37.75.dna.toplevel; Ensembl GRCh37 116 Homo_sapiens.GRCh37.dna.toplevel
gencode-grch37-p13	GRCh37	GENCODE 19 GRCh37.p13.genome
grch37-full	GRCh37	GRC GCA_000001405.14_GRCh37.p13_full_analysis_set
grch37-no-alt	GRCh37	GRC GCA_000001405.14_GRCh37.p13_no_alt_analysis_set
hg19	GRCh37	UCSC hg19
hg19-chromosomes	GRCh37	Illumina iGenomes UCSC/hg19
hg19-p13-full-analysis-set	GRCh37	UCSC hg19.p13.plusMT.full_analysis_set
hg19-p13-no-alt-analysis-set	GRCh37	UCSC hg19.p13.plusMT.no_alt_analysis_set
hg19-p13-plusmt	GRCh37	UCSC hg19.p13.plusMT
hs37d5	GRCh37	1000 Genomes hs37d5; Illumina iGenomes GATK/GRCh37
ncbi-build37.1	GRCh37	Illumina iGenomes NCBI/build37.1
ncbi-build37.2	GRCh37	Illumina iGenomes NCBI/build37.2
ensembl-grch38-primary	GRCh38	Ensembl 116 Homo_sapiens.GRCh38.dna.primary_assembly
ensembl-grch38-toplevel	GRCh38	Ensembl 116 Homo_sapiens.GRCh38.dna.toplevel
gencode-grch38-p14	GRCh38	GENCODE 50 GRCh38.p14.genome
gencode-grch38-primary	GRCh38	GENCODE 50 GRCh38.primary_assembly.genome
grch38-full	GRCh38	GRC GCA_000001405.15_GRCh38_full_analysis_set
grch38-full-hs38d1	GRCh38	GRC GCA_000001405.15_GRCh38_full_plus_hs38d1_analysis_set
grch38-no-alt	GRCh38	GRC GCA_000001405.15_GRCh38_no_alt_analysis_set; Illumina iGenomes NCBI/GRCh38
grch38-no-alt-hs38d1	GRCh38	GRC GCA_000001405.15_GRCh38_no_alt_plus_hs38d1_analysis_set
grch38-no-alt-hs38d1-igenomes	GRCh38	Illumina iGenomes NCBI/GRCh38Decoy
hg38	GRCh38	UCSC hg38
hg38-analysis-set	GRCh38	Illumina iGenomes UCSC/hg38; UCSC hg38.analysisSet
hg38-p14	GRCh38	UCSC hg38.p14
hs38dh	GRCh38	1000 Genomes GRCh38_full_analysis_set_plus_decoy_hla; Broad Institute Homo_sapiens_assembly38; Illumina iGenomes GATK/GRCh38
chm13v1.0	T2T-CHM13	T2T consortium chm13.draft_v1.0
chm13v1.1	T2T-CHM13	T2T consortium chm13.draft_v1.1; T2T consortium chm13v2.0_noY
chm13v2.0	T2T-CHM13	Illumina iGenomes UCSC/CHM13; T2T consortium chm13v2.0; UCSC hs1
chm13v2.0-dj-5s-rdna-phr-noy	T2T-CHM13	T2T consortium chm13v2.0_masked_DJ_5S_rDNA_PHR_noY
chm13v2.0-dj-5s-rdna-phr-noy-rcrs	T2T-CHM13	T2T consortium chm13v2.0_masked_DJ_5S_rDNA_PHR_noY_wi_rCRS
chm13v2.0-dj-5s-rdna-phr-xy	T2T-CHM13	T2T consortium chm13v2.0_masked_DJ_5S_rDNA_PHR_hardMaskY
chm13v2.0-dj-5s-rdna-phr-xy-rcrs	T2T-CHM13	T2T consortium chm13v2.0_masked_DJ_5S_rDNA_PHR_PAR_wi_rCRS
chm13v2.0-dj-rdna-phr-xx	T2T-CHM13	T2T consortium chm13v2.masked_DJ_rDNA_PHR.forXX
chm13v2.0-dj-rdna-phr-xy	T2T-CHM13	T2T consortium chm13v2.masked_DJ_rDNA_PHR.forXY
chm13v2.0-maskedY	T2T-CHM13	Broad Institute chm13v2.0.maskedY; T2T consortium chm13v2.0_maskedY
chm13v2.0-maskedY-rCRS	T2T-CHM13	T2T consortium chm13v2.0_maskedY_rCRS
chm13v2.0-maskedY-rCRS-EBV	T2T-CHM13	Broad Institute chm13v2.0.maskedY.rCRS.EBV
chm13v2.0-refseq	T2T-CHM13	NCBI RefSeq GCF_009914755.1_T2T-CHM13v2.0
ensembl-chm13v2.0	T2T-CHM13	Ensembl rapid release Homo_sapiens-GCA_009914755.4-unmasked

What is in it

All coordinates are 0-based, half-open. Sequence digests follow GA4GH refget (MD5 and SQ. sha512t24u); collection digests follow GA4GH seqcol.

File	Contents
roots.parquet	The 3,417 roots, with length and refget digests: every sequence of every release of each assembly (withdrawn patches included), and those editions add from outside it: the hs38d1 decoys, EBV, hg19's Yoruba mitochondrion and the hs37d5 decoy
roots_external.parquet	IMGT/HLA 3.18.0 alleles, by identifier and digest only (see below)
root_blocks.parquet	The bases of every root, uppercase with IUPAC codes kept, in 64 kb blocks, laid out for HTTP range reads
assembly_members.parquet	Every row of every assembly report each book draws on, with its book
aliases.parquet	Names of each root in the GenBank, RefSeq, UCSC, GRC, Ensembl and IMGT namespaces; a name that means something only within an assembly (chrM is the Yoruba sequence in hg19, rCRS in hg38) carries its book
layer_*.parquet	Annotations in root coordinates: AGP components and gaps, alt/patch placements and the padding they give (by book: a scaffold can be placed in two assemblies), named regions (centromeres, PARs), chrY's PARs, the GRC hard-masks, IUPAC positions, N runs, NCBI soft-masking, and the masks read from published masked files where no file gives them (T2T-CHM13's acrocentric and 5S rDNA masks, named by the regions T2T-Ref's README lists)
sequences.parquet	The sequence store: every distinct sequence (each root, and each derivative: masked, IUPAC→N, padded or oriented), by SQ. digest and MD5, with the root and nucleotide operations it is made from (and, for a padded or oriented one, the book whose placement it follows)
derivations.parquet	Every known way to make each sequence, and the editions that use it
editions.parquet	Each edition: book, label, identifier (an order-independent digest of its named sequences), seqcol digest, recipe, and facets (contents, operations, naming)
manifests.parquet	Each edition's sequences: name, root, operations, length, MD5, SQ. digest
targets.parquet	Each published file: its edition, publisher, name, URL and seqcol digest
target_sequences.parquet	Each published file's order of sequences
collections/*.json	Seqcol (level 2) documents for every edition and published file, named by digest
evidence/	What each published file was observed to contain, the source locks, and the root verification reports
catalog.json	The release manifest: every file's size, sha256 and role

Four nucleotide operations make every derivative. Two write N and keep coordinates: a mask (a layer's spans: the GRC hard-mask, chrY's PARs) and IUPAC→N. Two lay an alt scaffold or patch out by its placement on its parent: pad writes it in the parent's coordinates, N before and after, as Ensembl's GRCh37 files do; orient only reverse-complements one placed on the minus strand, as GENCODE 19 does.

Read any table directly with DuckDB, Polars or pyarrow, for example:

import polars as pl
base = "https://huggingface.co/datasets/abdenlab/refbook/resolve/v0.2.0/"
m = pl.read_parquet(base + "manifests.parquet")
m.filter(pl.col("name") == "chr21").select("edition", "md5", "operations").unique("md5")

The refbook Python library (github.com/abdenlab/refbook) identifies a header against this catalog, and resolves, materializes and fetches any collection whose sequences it holds.

Sources and licensing

The catalog tables are released under CC-BY-4.0. The underlying sequences are INSDC records (GenBank), distributed by NCBI without restrictions on use; the one root with no INSDC record, the hs37d5 decoy (1000G:hs37d5), is 1000 Genomes' concatenation of segments of other sequences, with its make-up as its AGP. The published files used as reproduction targets belong to their publishers and are not redistributed here; the evidence/ tables record only their sequence names, lengths and digests.

IMGT/HLA alleles are not redistributed. The IPD-IMGT/HLA Database is licensed under CC BY-NoDerivs, and its maintainers ask not to be mirrored. This catalog records only allele identifiers, names and digests, with links to the IMGT/HLA release. If you use the HLA entries, please cite IPD-IMGT/HLA (Barker et al., Nucleic Acids Research 2026; Robinson et al., Tissue Antigens 2000).

Versions

Each release is a tag (v0.2.0, …). Pin URLs to a tag, for example https://huggingface.co/datasets/abdenlab/refbook/resolve/v0.2.0/catalog.json.