Open adowneywall opened 5 years ago
Update:
This issue appears to be in the gene annotation file (.gtf
) file from NCBI. Upon initial examination there are 6505 transcripts belonging to 677 genes where the transcript id
is labeled as a gene.
Top ten examples:
3196 NC_035780.1 Gnomon transcript 2215606 2223571 . - . transcript_id gene115; gene_id gene115; gene_name LOC111129349;
3197 NC_035780.1 Gnomon exon 2215606 2215998 . - . transcript_id gene115; gene_id gene115; gene_name LOC111129349;
3198 NC_035780.1 Gnomon exon 2216590 2216764 . - . transcript_id gene115; gene_id gene115; gene_name LOC111129349;
3199 NC_035780.1 Gnomon exon 2216978 2217220 . - . transcript_id gene115; gene_id gene115; gene_name LOC111129349;
3200 NC_035780.1 Gnomon exon 2218775 2218835 . - . transcript_id gene115; gene_id gene115; gene_name LOC111129349;
3201 NC_035780.1 Gnomon exon 2221564 2221709 . - . transcript_id gene115; gene_id gene115; gene_name LOC111129349;
3202 NC_035780.1 Gnomon exon 2221868 2222168 . - . transcript_id gene115; gene_id gene115; gene_name LOC111129349;
3203 NC_035780.1 Gnomon exon 2222443 2222559 . - . transcript_id gene115; gene_id gene115; gene_name LOC111129349;
3204 NC_035780.1 Gnomon exon 2222660 2223211 . - . transcript_id gene115; gene_id gene115; gene_name LOC111129349;
3205 NC_035780.1 Gnomon exon 2223565 2223571 . - . transcript_id gene115; gene_id gene115; gene_name LOC111129349;
I created RSEM index based on the eastern oyster genome gtf file where I only considered transcripts based on the GNOMON source (this only appears to <100 transcripts that were from an RNAseq source, but were created problems with downstream analysis due to some issues with how they were labelled). I used the
rsem-prepare-reference
command with the following code:The issue appears to be the
*transcripts.fa
file generated identifies transcripts that have names like :gene9989_gene9989
, when I would expect something likerna16911_gene9988
. The appear to be a couple hundred of these erroneous transcripts within the.fa
file, and they appear to consistently have the duplicated gene ID pattern in the name (i.e.gene100_gene100
).This created some problems downstream in
salmon quant
step, which mapped non-trivial numbers of reads to those transcripts. I am still checking to confirm this issue did not also cause problems with the RSEM quantification.I'll be looking into what is creating this issue and if it is a command problem or an issue with the reference files we used.