Error occuring when ncfp encounters sequences which have been removed from NCBI database

[ArtingBioinfo]

Summary: NCBI sequence identifiers which have been removed from the NCBI data base cause a "no link/record returned for: xyz".

Description:

When attempting to use the ncfp command to create a file containing the nucleotide sequences from amino acid sequences using .fasta files which contain both the amino acid sequence and its associated accession number, a set of errors are listed when the command attempts to obtain the nucleotide sequences. The first being an index error: list index out of range and the second being: NCFPEFetchException: no link / record returned for: xyz

Reproducible Steps:

The command written which achieved these errors is as follows:
ncfp SigR_500_aa_seqs.fasta \ ncfp_nucleo_seqs \ leealexanderkeenan@gmail.com

Current Output:

Process input sequences: 100%|██████████████████████████████████████████████████████████████████████| 500/500 [04:14<00:00, 1.96it/s]
Search NT IDs: 1%|▍ | 3/500 [00:05<16:09, 1.95s/it]
Traceback (most recent call last):
File "/home/lee/miniconda3/lib/python3.9/site-packages/ncbi_cds_from_protein/entrez.py", line 216, in search_nt_ids
idlist = [lid["Id"] for lid in result[0]["LinkSetDb"][0]["Link"]]
IndexError: list index out of range

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "/home/lee/miniconda3/bin/ncfp", line 10, in
sys.exit(run_main())
File "/home/lee/miniconda3/lib/python3.9/site-packages/ncbi_cds_from_protein/scripts/ncfp.py", line 246, in run_main
addedrows, countfail = search_nt_ids(qrecords, cachepath, args.retries, disabletqdm=args.disabletqdm)
File "/home/lee/miniconda3/lib/python3.9/site-packages/ncbi_cds_from_protein/entrez.py", line 218, in search_nt_ids
raise NCFPEFetchException("No link/record returned for %s" % record.id)
ncbi_cds_from_protein.entrez.NCFPEFetchException: No link/record returned for WP_078606386.1

Expected Output:

Fasta file containing the nucleotide sequences returned from the amino acid sequences given.

Operating System: Linux (Linux for Windows, flavor Xubuntu)

[widdowquinn]

Thanks @ArtingBioinfo - could you please provide an input sequence file that could help me reproduce this error, so I can try to fix it?

[ArtingBioinfo]

Hello! I've attached the input file which was used to obtain the aforementioned error, I hope it helps! Thank you for your time and all the best, Lee Keenan

…

On Wed, 20 Jul 2022 at 16:02, Leighton Pritchard ***@***.***> wrote: Thanks @ArtingBioinfo <https://github.com/ArtingBioinfo> - could you please provide an input sequence file that could help me reproduce this error, so I can try to fix it? — Reply to this email directly, view it on GitHub <#34 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/A2BLZJMEDAIY3AUUSUZMPZDVVAIIJANCNFSM53PIADRQ> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[widdowquinn]

Hi @ArtingBioinfo - I think you'll have to attach the file through the web interface at #34 - the email attachment doesn't seem to come through.

L.

[widdowquinn]

Please don't provide sequences as a pdf. Please upload (drag/drop into the box) the FASTA file itself.

NOTE as GitHub doesn't recognise FASTA as a format, you can either compress the file as a .zip file, or change the file extension to .txt.

[ArtingBioinfo]

Galaxy26-[SigR_500_blastdbcmd] fasta.txt

That makes far more sense as to why I couldn't originally upload the file. That should be an acceptable txt file now.

Apologies for the inconvenience caused.

[widdowquinn]

The issue seems to arise from input sequences that have been suppressed or removed in NCBI, such as

• WP_078627170.1 search
• WP_078627170.1 suppressed record

For sequences such as these, which are not annotated on any genome, we will not be able to recover a coding sequence.

In addition, these WP_ records are identical protein groups (IPGs). This is the way that RefSeq keeps database size small - it bundles all identical/non-redundant proteins together in a single record. Although the protein sequences may be identical, the underlying coding sequences for each individual protein might not be, and ncfp does not currently do the job of tracking all of those underlying sequences and assigning new IDs so that they can be backtranslated. It is better to remove all the MULTISPECIES/RefSeq IPGs from your input.