Difference between revisions of "Generating Initial GAF file for Upload to Postgres"

From WormBaseWiki
Jump to navigationJump to search
Line 12: Line 12:
  
 
= Notes =
 
= Notes =
−
*The assumption here is that if a paper-gene connection exists from manual annotation, that it will have covered all of the possible annotations from that paper.  This may not always be 100% true, but it will be simpler to filter this way than to try to also check the actual term used and determine parent-child (i.e. more or less granular) relations to decide which annotations to keep.  Typically, the manual annotations are more granular than the Phenotype2GO annotations.
+
*The assumption here is that if a paper-gene connection exists from manual annotation, that it will have covered all of the possible annotations from that paper.  This may not always be 100% true but, at least for now, it will be simpler to filter this way than to also check the actual term used and determine parent-child (i.e. more or less granular) relations to decide which annotations to keep.  Typically, the manual annotations are more granular than the Phenotype2GO annotations anyway.  We can always re-visit this assumption, if needed.
  
  
  
 
Back to [[20141022_-_Phenotype2GO_Pipeline]]
 
Back to [[20141022_-_Phenotype2GO_Pipeline]]

Revision as of 15:38, 5 November 2014

Initial Round of Entering Phenotype2GO-Based Annotations into Postgres

  • The idea here is to generate a non-redundant set of Phenotype2GO annotations to enter into postgres. Non-redundant means that these annotations do no overlap with any existing manual annotations.
  • Step 1: Retrieve the phenotype2go.wsxxx.wb file from the ftp site
  • Step 2: Using the annotations retrieved from Step 1, create a two-column table that contains the WBGene ID found in Column 2 and the PMID value found in Column 6.
  • Step 3: Using the gp_association file from UniProt-GOA, first map Column 2 value (a UniProtKB ID) to a WBGene ID using the WB gp2protein file and then replace the UniProtKB ID in Column 2 with the corresponding WBGene ID.
  • Step 4: Output a list of any UniProtKB IDs that don't map to a WBGene ID and update the gp2protein file.
  • Step 5: Repeat Step 3 with an updated gp2protein file, if needed.
  • Step 6: Create a second two-column table that contains the WBGene ID now found in Column 2 of the gp_association file and the PMID value found in Column 5 of the gp_association file.
  • Step 7: Compare the values in each of the two tables and then output two files from the initial file of Phenotype2GO annotations: 1) a file containing those annotations that are redundant with the gp_association annotations (i.e., there is an exact match between the gene and the paper in each table), and 2) a file containing those annotations that are NOT present in the gp_association file (i.e., the gene-paper combination exists in the Phenotype2GO file but NOT in the gp_association file).
  • Step 8: Upload the annotations in file #2 (non-redundant) to postgres gop_ OA tables.

Notes

  • The assumption here is that if a paper-gene connection exists from manual annotation, that it will have covered all of the possible annotations from that paper. This may not always be 100% true but, at least for now, it will be simpler to filter this way than to also check the actual term used and determine parent-child (i.e. more or less granular) relations to decide which annotations to keep. Typically, the manual annotations are more granular than the Phenotype2GO annotations anyway. We can always re-visit this assumption, if needed.


Back to 20141022_-_Phenotype2GO_Pipeline

Retrieved from "https://wiki.wormbase.org/index.php?title=Generating_Initial_GAF_file_for_Upload_to_Postgres&oldid=25299"