Hi. I used RNA-seq and protein data to annotate the genes and I want to assign the gene ontology terms to the genes. To minimize duplications, I obtained the longest isoform with agat and then detected orthogroups/orthologs with Orthofinder. InterproScan, eggnog-mapper, and deepGoplus failed in my HPC; but, there are related species in NCBI with gene_ontology files. I used the related species. I want to transfer the go-terms, either pairwisely, but I thought it would be more interesting to see - as confidence, how many go-terms are supported by multiple-species. The orthogroups file looks like this:
head Orthogroups.tsv
Orthogroup Canlup Hydlep LW1 Lepwed Mirang Mirleo Neosch Ursarc
where each column is separated by tab, and each gene within a column by ", ".
The GOA files look like this:
head GCF_002201575.2_ASM220157v2_gene_ontology.gaf -n 12
!gaf-version: 2.2
!generated-by: NCBI
!date-generated: 2023-11-10
!
!interProScan version: 5.64-96.0
!PANTHER version: 17.0
!GO version: 2023-07-27
!
!#DB GeneID Symbol Qualifier GO_ID Reference Evidence_Code With,From Aspect Gene_Name Gene_Synonym Type Taxon Date Assigned_By Annot_Ext Gene_Product_Form_ID
NCBIGene 110569606 SHROOM2 located_in GO:0005912 PMID:30032202 IEA PANTHER:PTN001805876|UniProtKB:A0A5F4BXY6 C NA protein taxon:29088 20231110 RefSeq
NCBIGene 110569606 SHROOM2 involved_in GO:0007015 PMID:30032202 IEA PANTHER:PTN001805876|UniProtKB:A0A5F4BXY6 P NA protein taxon:29088 20231110 RefSeq
NCBIGene 110569606 SHROOM2 located_in GO:0016324 PMID:30032202 IEA PANTHER:PTN001805876|UniProtKB:A0A5F4BXY6 C NA protein taxon:29088 20231110 RefSeq
I extracted the gene ontology terms with:
awk '!/^!/ {print $2 "\t" $5}' species.gaf > species.gene2go.tsv
and now I have the gene_id<TAB>GO:XXXXXXX table
head Neosch.gene2go.tsv
110569606 GO:0005912
110569606 GO:0007015
110569606 GO:0016324
110569606 GO:0030864
110569606 GO:0043296
110569606 GO:0051015
How can I map/transfer the go-terms to the genes?
Thanks;
0 answers
No answers yet.
Log in to answer this question.
I am not sure that I understood the questions correctly, but if you have orthologs then you need the geneid1 to geneid2 mapping then you map your geneid to an annotated one, and make a new goa file that only contains those of your new gene IDs that had a match
This is concerning because they can be CPU intensive but this shouldn't be an issue if you have HPC access. Why did they fail? You can chunk your input fasta into manageable sizes (e.g., 1000 sequences each) and submit your jobs across an array, then merge the final output into a single file.
Can you elaborate here? I don't understand the question. GO terms are designed to be generalised and species agnostic if that was the question.