Human Feature to Gene Conversion Table

A data.table used to convert from feature IDs to various identifiers.

Usage

HUMAN_FEATURE_TO_GENE

Format

A sorted data.table with 1,920,618 rows and 12 columns:

assay

character; the assay (ome). One of "epigen-atac-seq", "epigen-methylcap-seq", "metab", "prot-clinical", "prot-ol", "prot-ph", "prot-pr", or "transcript-rna-seq".

feature_id

factor; the feature identifier.

entrez_gene

factor; Entrez gene identifier.

gene_symbol

factor; gene symbol.

ensembl_gene

factor; ensembl gene identifier.

uniprot

factor; UniProt identifier.

refmet_name

factor; RefMet metabolite name, from the pinned RefMet snapshot.

refmet_id

factor; RefMet metabolite identifier, from the pinned RefMet snapshot.

kegg_id

factor; KEGG identifier, from the pinned RefMet/KEGG snapshot.

custom_annotation

factor; the region of the assigned gene the peak falls in — one of "Promoter (<=1kb)", "Promoter (1-2kb)", "5' UTR", "3' UTR", "Exon", "Intron", "Overlaps Gene", "Upstream (<5kb)", "Downstream (<5kb)" or "Distal Intergenic". NA unless assay is "epigen-atac-seq" or "epigen-methylcap-seq".

relationship_to_gene

numeric; the signed distance in base pairs from the peak to the assigned gene, 0 where the peak overlaps it. NA unless assay is "epigen-atac-seq" or "epigen-methylcap-seq".

flanking_sequence

factor; flanking sequence. Only applicable if assay is "prot-ph".

Source

Built by Stage 1 step 07 of the motrpac-human-presuspension-repro pipeline. The refmet_name, refmet_id and kegg_id columns come from a pinned offline RefMet/KEGG snapshot rather than a live Metabolomics Workbench query, so the mapping does not move with those databases. The peak annotations in custom_annotation and relationship_to_gene are produced by ChIPseeker against the pinned Ensembl v105 TxDb.

Details

custom_annotation and relationship_to_gene describe where an ATAC-seq or MethylCap-seq peak sits relative to the gene it was assigned to. Without them a peak in a promoter and a peak 40 kb into an intron are indistinguishable once mapped, since both carry only the gene. Both are derived from the peak coordinates in the feature_id rather than measured per tissue, so they take the same value in every tissue a peak appears in.

Per-tissue measurements are deliberately not carried here. This table is keyed on (assay, feature_id) with no tissue column, so it could only hold a collapse across tissues. The phosphosite localization flag (confident_site) is the case in point: it was added in 2.0.3 and removed in 2.0.7 because muscle and adipose disagree on 859 of their 7,865 shared prot-ph sites. Read it per tissue from *_PROT_PH_QC$feature_metadata in MotrpacHumanPreSuspensionData.