added --min_qual parameter

[MarieLataretu]

• sets the minimum variant quality for consensus masking in the ARTIC workflow
• (ARTIC) default is 20

I run the pipeline locally without errors and expected outcome changes.

[DataSpott]

Testing it right now. Looking at the code, I guess we should also apply this to "artic_medaka_custom_bed"!?

[MarieLataretu]

Testing it right now. Looking at the code, I guess we should also apply this to "artic_medaka_custom_bed"!?

Upsi, forgot this in the heat of the moment. I'm on it

[DataSpott]

Tested it now with two datasets:

• 1st: 12 fastqs from our routine-seq with varying quality and lineages. These were formerly run through poreCov and thereby already filtered. Tested the default V1200 setting as well as the custom-bed with them.
• 2nd: 4 "raw" fastq-files from wastewater sequencing, which I run with the custom-bed.

Workflow runs through, perfectly fine. Yet, the results are exactly the same between the master and this branch. So, I am unsure whether the parameter-flag does not work as intended or whether my datasets are not suited for testing it!?
Do you see any difference in your test runs, and if so, in which metric (I would expect a shift in the number of "N"s as it affects consensus masking)?

[MarieLataretu]

Probably you don't have variants with variant qc below --min_qual (last value in the VCF file)?

E.g.: (results/3.Lineages_Clades_Mutations/sample/sample.pass.vcf)

# --min_qual 20 (default)
MN908947.3	22030	.	GT	G	23.925	PASS	DP=1200;DPS=600,600;Pool=1	GT:GQ	1:24
MN908947.3	22107	.	A	G	34.108	PASS	DP=1200;DPS=600,600;Pool=1	GT:GQ	1:34

vs

# --min_qual 19
MN908947.3	22030	.	GT	G	23.925	PASS	DP=1200;DPS=600,600;Pool=1	GT:GQ	1:24
MN908947.3	22034	.	A	AG	19.447	PASS	DP=1200;DPS=600,600;Pool=1	GT:GQ	1:19
MN908947.3	22107	.	A	G	34.108	PASS	DP=1200;DPS=600,600;Pool=1	GT:GQ	1:34

The --min_qual 20 example causes a frameshift in the spike; for SNP's one should see +/- N in the consensus.
(I can't really explain why we are seeing this; 22034 used to have higher variant qc values.)

[DataSpott]

Screened now through all the pass.vcf-files of my test runs and, as you said, couldn't find a single case of too low quality with the new setting. Some entries were 1:20 (quality: 20.XXX), just enough to get accepted, but nothing below.

How did you come to the cut-off of 20? Looking into the (quite outdated) artic manual at https://artic.readthedocs.io/en/latest/commands/, this specific flag isn't even specified...? Only [--min-mapq], which reads slightly different, but had a default value of 20.

[MarieLataretu]

How did you come to the cut-off of 20? Looking into the (quite outdated) artic manual at https://artic.readthedocs.io/en/latest/commands/, this specific flag isn't even specified...? Only [--min-mapq], which reads slightly different, but had a default value of 20.

Great, on the CLI the default value is not printed... I checked the following:
a) the code here, and
b) the minion.log.txt in an ARTIC work dir says artic_vcf_filter --medaka --min-depth 20 --min-qual 20

[MarieLataretu]

How did you come to the cut-off of 20? Looking into the (quite outdated) artic manual at https://artic.readthedocs.io/en/latest/commands/, this specific flag isn't even specified...? Only [--min-mapq], which reads slightly different, but had a default value of 20.

Great, on the CLI the default value is not printed, either way, I checked the following:
a) the code here, and
b) the minion.log.txt in an ARTIC work dir says artic_vcf_filter --medaka --min-depth 20 --min-qual 20

[DataSpott]

Alright, so artic already applied this parameter in the background with its default value of 20. Then the results are reproducible if the parameter is unchanged. Workflow is working. Fine on my end, I will merge it into the master:)