Mapping of Compound IDs

[joewandy]

Sometimes a compound is mapped by ID to some ChEBI ID, but it turns out in Reactome a different ID is being used to relate the compound to reactions and pathways. This leads to some compounds missing from the table. We need to fix this.

e.g. for Histidine, often people use this ID to identify the compound

https://www.ebi.ac.uk/chebi/searchId.do?chebiId=CHEBI:27570

but Reactome uses this ID, which is the zwitterion form:

https://www.ebi.ac.uk/chebi/searchId.do?chebiId=CHEBI:57595

[RonanDaly]

Have some R code to do this along with a mapping from ChEBI.

[RonanDaly]

Here's the R. It looks like I was using the mapping to construct a graph and performing a breadth first search to find linked entities. This might not be a good way to do it!

expandChEBIIds = function(ids) {
  chebi_relation_file = system.file("extdata", "relation.tsv", package = "polydatar")
  df = read.table(chebi_relation_file, stringsAsFactors = FALSE, sep='\t', header = TRUE)
  mapping = df[df$TYPE == 'is_conjugate_acid_of' | df$TYPE == 'is_conjugate_base_of' | df$TYPE == 'is_tautomer_of',c('INIT_ID', 'FINAL_ID')]
  mapping$INIT_ID = as.character(mapping$INIT_ID)
  mapping$FINAL_ID = as.character(mapping$FINAL_ID)

  vertices = unique(c(mapping$INIT_ID, mapping$FINAL_ID, ids))
  g = graph_from_data_frame(mapping, vertices=vertices)
  map = as.list(ids)
  names(map) = ids

  for ( i in seq_along(map) ) {
    b = bfs(g, root=ids[i], unreachable=FALSE, order=TRUE)
    map[[i]] = names(b$order[!is.na(b$order)])
  }
  return(map)
}

[RonanDaly]

The file relation.tsv was downloaded from ChEBI, ftp://ftp.ebi.ac.uk/pub/databases/chebi/Flat_file_tab_delimited/

[joewandy]

The codes is being implemented into pyMultiOmics in this issue glasgowcompbio/pyMultiOmics#5.

Once pyMultiOmics is integrated into this project as part of #80, we can close this issue.