UCSC databases (TxDB)

[ivirshup]

Description of feature

Getting UCSC data from TxDB bioconductor sources

[ivirshup]

It would be useful to write down what exactly the differences between EnsDB and TxDB are.

To play around with this:

access via ibis

import genomic_features as gf
import ibis

!wget https://bioconductorhubs.blob.core.windows.net/annotationhub/ucsc/standard/3.15/TxDb.Hsapiens.UCSC.hg38.knownGene.sqlite

ensdb = gf.ensembl.annotation(species="Hsapiens", version="108").db
ucscdb = ibis.connect("TxDb.Hsapiens.UCSC.hg38.knownGene.sqlite")

for tbl_name in ensdb.list_tables():
    print(tbl_name, ensdb.table(tbl_name).schema())

EnsDB schema

chromosome ibis.Schema {
  seq_name     string
  seq_length   int32
  is_circular  int32
}
entrezgene ibis.Schema {
  gene_id   string
  entrezid  int32
}
exon ibis.Schema {
  exon_id         string
  exon_seq_start  int32
  exon_seq_end    int32
}
gene ibis.Schema {
  gene_id               string
  gene_name             string
  gene_biotype          string
  gene_seq_start        int32
  gene_seq_end          int32
  seq_name              string
  seq_strand            int32
  seq_coord_system      string
  description           string
  gene_id_version       string
  canonical_transcript  string
}
metadata ibis.Schema {
  name   string
  value  string
}
protein ibis.Schema {
  tx_id             string
  protein_id        string
  protein_sequence  string
}
protein_domain ibis.Schema {
  protein_id             string
  protein_domain_id      string
  protein_domain_source  string
  interpro_accession     string
  prot_dom_start         int32
  prot_dom_end           int32
}
tx ibis.Schema {
  tx_id             string
  tx_biotype        string
  tx_seq_start      int32
  tx_seq_end        int32
  tx_cds_seq_start  int32
  tx_cds_seq_end    int32
  gene_id           string
  tx_support_level  int32
  tx_id_version     string
  gc_content        float64
  tx_external_name  string
  tx_is_canonical   int32
}
tx2exon ibis.Schema {
  tx_id     string
  exon_id   string
  exon_idx  int32
}
uniprot ibis.Schema {
  protein_id            string
  uniprot_id            string
  uniprot_db            string
  uniprot_mapping_type  string
}

for tbl_name in ucscdb.list_tables():
    print(tbl_name, ucscdb.table(tbl_name).schema())

TxDB schema

cds ibis.Schema {
  _cds_id     int32
  cds_name    string
  cds_chrom   !string
  cds_strand  !string
  cds_start   !int32
  cds_end     !int32
}
chrominfo ibis.Schema {
  _chrom_id    int32
  chrom        !string
  length       int32
  is_circular  int32
}
exon ibis.Schema {
  _exon_id     int32
  exon_name    string
  exon_chrom   !string
  exon_strand  !string
  exon_start   !int32
  exon_end     !int32
}
gene ibis.Schema {
  gene_id  !string
  _tx_id   !int32
}
metadata ibis.Schema {
  name   string
  value  string
}
splicing ibis.Schema {
  _tx_id     !int32
  exon_rank  !int32
  _exon_id   !int32
  _cds_id    int32
  cds_phase  int32
}
transcript ibis.Schema {
  _tx_id     int32
  tx_name    string
  tx_type    string
  tx_chrom   !string
  tx_strand  !string
  tx_start   !int32
  tx_end     !int32
}

It does look like the UCSC sqlite files carry less information.

Docs/ links

• https://bioconductor.org/packages/release/bioc/html/GenomicFeatures.html
  • https://bioconductor.org/packages/release/bioc/vignettes/GenomicFeatures/inst/doc/GenomicFeatures.html
• https://bioconductor.org/packages/release/bioc/html/ensembldb.html
• Common filters are defined here: https://bioconductor.org/packages/release/bioc/vignettes/AnnotationFilter/inst/doc/AnnotationFilter.html
• https://github.com/Bioconductor/GenomeInfoDb for interop

It's probably worth looking into how the bioconductor packages deal with having two different schemas. E.g. do they subclass, are the annotation filters aware?

cc: @nvictus

[ivirshup]

Re discussion about nonstandard chromosome names @nvictus: jorainer/ensembldb#88