Removing Dead Genes from Papers
Removing Dead Genes: Transferring Data from Old to New Postgres Tables
When populating the new paper tables in postgres, we realized that if a gene paper connection had different evidences and one evidence was valid while another was invalid, the valid gene-paper connection was transferred even if the invalid connection was actually correct.
This resulted in a resurfacing of dead gene-paper connections for the WS213 build.
To fix this, we:
1) Created a new postgres table that contains information from WB on dead genes, i.e. the status of the gene and the action performed, e.g. mergrd or just dead.
2) For simple merges, we replaced the dead gene-paper connection with a corresponding live gene-paper connection, using the WBGene ID of the gene that acquired the merge.
3) For dead genes with no merge, we simply removed the gene-paper connection.
4) One more complicated gene model change that involved mergers and splits was handled manually before transferring data to the new tables.