BioMart offline usage

[zhenzuo2]

Description of feature

Hi,

Thank you for developing epitopeprediction! I got error when running at line 1133

transcriptProteinTable = ma.get_protein_ids_from_transcripts(transcripts, type=EIdentifierTypes.ENSEMBL).

If the input variant list is too long and then BioMart will decline my request (due to too many times). Is there a way I can run it locally? Thank you!

Best,

Zhen

[jonasscheid]

Hi! Thanks for using the Pipeline.

Currently, parsing of a local biomart version is not possible. We rely on querying biomart unfortunately

[jonasscheid]

Solutions to use a local biomart version might be a good addition to the pipeline, e.g. implementing pyensemble in the variant prediction part

[zhenzuo2]

Thank you so much for your prompt response!

[christopher-mohr]

Description of feature

Hi,

Thank you for developing epitopeprediction! I got error when running at line 1133

transcriptProteinTable = ma.get_protein_ids_from_transcripts(transcripts, type=EIdentifierTypes.ENSEMBL).

If the input variant list is too long and then BioMart will decline my request (due to too many times). Is there a way I can run it locally? Thank you!

Best,

Zhen

Hi @zhenzuo2,

Did you try using the "splitting functionality" that is implemented in the pipeline? You can get an overview of the parameters that can be used here under "Run optimisation": https://nf-co.re/epitopeprediction/2.3.1/parameters/

Not sure if it would help in your case but it's worth a try.

Best,
Chris

[zhenzuo2]

Description of feature

Hi,
Thank you for developing epitopeprediction! I got error when running at line 1133
transcriptProteinTable = ma.get_protein_ids_from_transcripts(transcripts, type=EIdentifierTypes.ENSEMBL).
If the input variant list is too long and then BioMart will decline my request (due to too many times). Is there a way I can run it locally? Thank you!
Best,
Zhen

Hi @zhenzuo2,

Did you try using the "splitting functionality" that is implemented in the pipeline? You can get an overview of the parameters that can be used here under "Run optimisation": https://nf-co.re/epitopeprediction/2.3.1/parameters/

Not sure if it would help in your case but it's worth a try.

Best, Chris

Description of feature

Hi,
Thank you for developing epitopeprediction! I got error when running at line 1133
transcriptProteinTable = ma.get_protein_ids_from_transcripts(transcripts, type=EIdentifierTypes.ENSEMBL).
If the input variant list is too long and then BioMart will decline my request (due to too many times). Is there a way I can run it locally? Thank you!
Best,
Zhen

Hi @zhenzuo2,

Did you try using the "splitting functionality" that is implemented in the pipeline? You can get an overview of the parameters that can be used here under "Run optimisation": https://nf-co.re/epitopeprediction/2.3.1/parameters/

Not sure if it would help in your case but it's worth a try.

Best, Chris

Thank you for sharing this, Chris. I haven’t had a chance to try it yet. The issue is that for security reason, the computing servers we use are not allowed to connect to the internet. My current solution is to download dataframe from Biomart and use as an input file to that function.

def get_protein_ids_from_transcripts_offline(transcripts, data_path = "mart_export.txt"):
    df = pd.read_csv(data_path)
    result = df.loc[df["Transcript stable ID version"].isin(transcripts),["Protein stable ID","RefSeq peptide ID","UniProtKB/Swiss-Prot ID", "Transcript stable ID version"]]
    result.columns = ["ensembl_id", "refseq_id",
                        "uniprot_id", "transcript_id"]
    print("Offline Now!") 
    return result

Using similar ways I changed a few other functions, such as generate_transcripts_from_variants() and generate_peptides_from_variants(). It works now. I will try "Run optimisation" you mentioned.

[jonasscheid]

We would be very happy to see an offiline / download implementation of this in the pipeline, since biomart servers can be quite unstable. If you have time to open a PR, that would be great @zhenzuo2!

[jonasscheid]

FYI @zhenzuo2 Thanks to @SusiJo there is now an implementation in the dev brach. Thank you both 🙏🏼