AA sequences ending with * have missing values for molecular_weight and hydrophobicity columns and inconsistent values on the CDS_stop_codon_found column in Ampcombi_summary.tsv

[amizeranschi]

While testing the AMP screening with Macrel, the Ampcombi_summary.tsv file shows several AA sequences ending with an asterisk, which I'm assuming represents a stop codon, as seen in a recent discussion from elsewhere: bcgsc/AMPlify#17

It appears that all the sequences ending with an asterisk in Ampcombi_summary.tsv have corresponding missing values on the molecular_weight and hydrophobicity columns. In addition, the CDS_stop_codon_found seems to have inconsistent values for these sequences, as some of them report a stop codon, while others do not.

This is the command that I ran:

nextflow run nf-core/funcscan -r add_interproscan_to_amp -profile docker,test_full --run_amp_screening --amp_skip_ampir --amp_skip_amplify --run_bgc_screening false --run_arg_screening false
--amp_ampcombi_parsetables_cutoff 0.4 --save_db --outdir test-funcscan

And attached is the Ampcombi_summary.tsv file:

Ampcombi_summary.tsv.txt

[jfy133]

This is being addressed in #447 I believe, correct @jasmezz ?

[jasmezz]

It's solved for Pyrodigal in that PR, yes. Maybe I should append a string filtering step for the other annotation tools to get rid of the asterisk, because it cannot be deactivated by Bakta, Prokka, Prodigal.

[amizeranschi]

In order to prevent issues with Ampcombi's report generation, it might be a good idea to suppress or eliminate stop codons from AA sequences whenever the pipeline is run with --run_amp_screening, regardless of which gene annotation and AMP screening tools are enabled.

[jasmezz]

Agree. I'll test whether it also has an influence on other screening workflows, but this check and * removal should probably be on pipeline level (not AMPcombi only).

[jfy133]

Can this be closed @jasmezz ?

[jasmezz]

Yes! @amizeranschi and I tested today that everything works as expected.