pubrunner.yml

name: FoodRelations
version: 0.1
url: https://github.com/jakelever/foodrelations
resources:
   all:
      - PUBCHEM
      - FOODON
   full:
      - PUBMED:
           format: bioc
      - AGRICOLA:
           format: bioc
   test:
      - PUBMED_TWOFILES:
           rename: PUBMED
           format: bioc
      - AGRICOLA_TWOFILES:
           rename: AGRICOLA
           format: bioc
build:
   - python generateDrugTerms.py --pubChemMeshPharmFile {IN:PUBCHEM}/MeSH-Pharm --stopwords stopwords.chemical.plusCommon.txt --outFile {OUT:chemical.json}
   - cp food.json {OUT:food.json}
run:
   - python cooccurrenceExtractor.py --biocFile {IN:AGRICOLA/%} --wordlist1 {IN:food.json} --wordlist2 {IN:chemical.json} --outFile {OUT:tmp/%.cooccurrences}
   - python cooccurrenceExtractor.py --biocFile {IN:PUBMED/%} --wordlist1 {IN:food.json} --wordlist2 {IN:chemical.json} --outFile {OUT:tmp/%.cooccurrences}
   - python combineCooccurrences.py --inDir {IN:tmp} --outFile {OUT:results.txt}
output: results.txt