Dear all, I would like see which variant candidates were selected by deepvariant, as well as edit its sensitivity parameters. From the documentation page, I see that I can do both things by adding the parameters: -

--debug_output_all_candidates=ALT

-

--make_examples_extra_args="vsc_min_count_snps=2,vsc_min_fraction_snps=0.12,vsc_min_count_indels=2,vsc_min_fraction_indels=0.06"

What I did: • I added

--debug_output_all_candidates=ALT

to _deepvariant_rundeepvariant_ (in

modules.config

), but this made the process crash. The reason is that the parameter is not recognized, since it is exclusive to the postprocess step. • The postprocess step is the 3rd of the 3 under-the-hood steps that implement deepvariant. So, to use the parameter, I replaced deepvariant_rundeepvariant with the 3 processes: deepvariant_makeexamples, deepvariant_callvariants, deepvariant_postprocessvariants, and placed the parameter in the _deepvariant_postprocessvariants_'

modules.config

. • I left _deepvariant_makeexamples_ and _deepvariant_callvariants_ configured with their defauls. However, deepvariant_makeexamples crashes:

I0107 09:26:26.150934 128396548341760 genomics_reader.py:223] Reading my-sample.bam with NativeSamReader
E0107 09:26:26.153316 128396548341760 errors.py:61] --channel_list is required. A good default value to use is `--channel_list=read_base,base_quality,mapping_quality,strand,read_supports_variant,base_differs_from_ref`.
  /tmp/Bazel.runfiles_476xicyy/runfiles/absl_py/absl/flags/_validators.py:233: UserWarning: Flag --mode has a non-None default value; therefore, mark_flag_as_required will pass even if flag is not specified in the command line!
    mark_flag_as_required(flag_name, flag_values)

Can it be that the error is raised because I left the input

gzi

channel of deepvariant_makeexamples empty? What even is

gzi

?

DEEPVARIANT_MAKEEXAMPLES(
    ch_deepvariant_in,  // <-- has the required bam/bai/bed input
    ch_fasta,           // <-- has the required reference sequence
    ch_fai,             // <-- has the required reference sequence index
    [[], []],           // <-- gzi
    [[], []]            // <-- par_bed
)

gzi would be the fasta gzi index I assume

Ah, got it, thank you. Then I don't think that's the reason why the process crashes.

Let me check how we're running deepvariant in sarek

Thank you, Maxime. Yes, you're only using rundeepvariant. Meanwhile I realized that I can use a nf-core's deepvariant subworkflow to call the three makeexamples/callvariants/postprocessvariants. I therefore replaced my solution with nf-core's subworkflow. However, I obtain the same error:

I0107 12:24:54.857266 125512379604992 genomics_reader.py:223] Reading my-sample.bam with NativeSamReader
  E0107 12:24:54.859075 125512379604992 errors.py:61] --channel_list is required. A good default value to use is `--channel_list=read_base,base_quality,mapping_quality,strand,read_supports_variant,base_differs_from_ref`.

🫠