QC - new GOA-GOC data exchange pipeline

[pgaudet]

The data coming from the GOA pipeline is on AmiGO staging. This ticket looks at the differences across the two datasets.

GOA ftp: https://ftp.ebi.ac.uk/pub/contrib/goa/panther_proteomes/ GOA error reports for external groups: https://ftp.ebi.ac.uk/pub/contrib/goa/reports/ Stats (GOC release): https://docs.google.com/spreadsheets/d/1asamlC32E8HDGCqUVaE1O3hp4jp-Y6Z7nt6PrK_6jGw/edit?gid=0#gid=0

Need to check all sources

• [ ] dictybase
• [ ] ecocyc >> alreaddy loaded via GOA; GOC only loads GPI
• [ ] fb
• [x] genedb
• [x] goa
• [ ] japonicusdb
• [ ] mgi
• [ ] noctua
• [ ] paint
• [ ] pombase
• [ ] pseudocap
• [ ] reactome
• [ ] rgd
• [x] rnacentral (GOC only loads GPI)
• [ ] sgd
• [ ] sgn
• [ ] tair
• [ ] wb
• [ ] xenbase
• [ ] zfin

Known differences between GOA and GOC pipeline:

• ND filtered by GOA
• interontology links are instantiated
• do_not_annotate are filetered

[pgaudet]

Decrease in ND, since GOA is filtering ND annotations if there are other annotations

[pgaudet]

pombe taxon is different between pombase ad Uniprot

[pgaudet]

SGD: AmiGO staging: 54,426 annotations assigned by SGD QuickGO: 54,733 annotation sassigned by SGD

S. cerevisiae: AmiGO staging: 117,992 annotations QuickGO: 146,440 annotations (RNA, complexes and SP Reviewed proteins)

[pgaudet]

Human data:

• Large increase in IEAs:
• • 73K human IEAs on amigo-current (protein only) 121k (all entity types); 235k amigo-staging (protein only) & 283k (all entity types); QuickGO: 1M; reference proteomes only = 418k