ENSEMBL fasta format parsing

[tsbatth]

Hi Vadim and team,

I have a question regarding parsing of ENSEMBL fasta files. I was wondering if there is a way for DIA-NN to parse the FASTA with the following format:

>ENSRNOP00000074671.1 pep chromosome:Rnor_6.0:10:45338062:45480999:-1 gene:ENSRNOG00000058068.1 transcript:ENSRNOT00000078353.1 gene_biotype:protein_coding transcript_biotype:protein_coding gene_symbol:Obscn description:obscurin, cytoskeletal calmodulin and titin-interacting RhoGEF [Source:RGD Symbol;Acc:631335]
MDHSFSGAPRFLTRPK....

Thanks for your time!

Kind regards,

Tanveer

[vdemichev]

Hi Tanveer,

It will get the ID (after '>') correctly, but not the rest of it.
Fairly easy to add support for gene names. A temporary workaround would be to replace "gene:" with "GN=" in the FASTA file.

Best,
Vadim

[tsbatth]

Thanks for the tip, I will try that out or we can just assemble the protein/names after.

You are correct, when searching with standard ENSEMBLE fasta, the raw file search it will give the following output which made it seem like there were no proteins identified (the 30 are from a contaminant fasta which has standard headers):

[918:49] 45399 library precursors are potentially detectable
[918:49] Processing...
[918:50] RT window set to 0.924791
[918:50] Recommended MS1 mass accuracy setting: 5.33463 ppm
[918:51] Removing low confidence identifications
[918:51] Removing interfering precursors
[918:52] Training neural networks: 40976 targets, 24245 decoys
[918:55] Number of IDs at 0.01 FDR: 34907
[918:58] Removing low confidence identifications
[918:58] Removing interfering precursors
[918:59] Training neural networks: 41356 targets, 24437 decoys
[919:02] Number of IDs at 0.01 FDR: 35160
[919:03] Calculating protein q-values
[919:03] Number of proteins identified at 1% FDR: 30 (precursor-level), 0 (protein-level) (inference performed using proteotypic peptides only)
[919:03] Quantification

However in the quantitative matrices there are still over 4000 protein groups annotated, despite the fact that protein names and gene names just says "pep" . So it seems to work despite it saying 0 protein level identifications in the DIA-NN console.

Regards,

Tanveer

[vdemichev]

Yes, all as expected. 0 protein-level IDs = 0 genes identified. To count isoform IDs can switch protein inference to isoforms.

[Amon789]

Hi Tanveer and Vadim,

I tried to change gene_symbol from the ensembl header to GN= but the DIA-NN outputs did not yet recognize the gene names and they are all pep.

Are there any efficient way to annotate these ensembl protein stable IDs with version( ENSRNOP00000074671.1 ) to gene names ? Ensembl Biomart is not efficient.

BR
Ahmed

[tsbatth]

Hey Ahmed,

I found deleting "pep" in all the FASTA headers seemed to fix the issue.

Regards,

T