Read-level STR sizing results

[MaestSi]

Dear all,
I'm running STRdust v0.20.0 on ONT amplicon sequencing data.
I am interested in obtaining the number of repeats for each read at the relevant locus.
I tried adding the --somatic parameter, but it only resulted in the addition of the SEQS field, which should correspond to "Sequences supporting the two alleles", according to the vcf header.
Is it possible to obtain read-level estimates?
Thanks in advance,
Simone

[wdecoster]

Yes, --somatic gives you the repeat sequence for each read, not the number of repeats per read. So you could take those sequences as repeat length, optionally divided by motif length. A bit of Python code would get the job done, would you need help with that? Note that at the individual read length level, there is some more technical sequencing noise.

[MaestSi]

Thanks for the information. I think I can manage to take those sequences and divide their length by the motif length. I was just surprised by the fact that the number of sequences I get with STRdust (ranging from 0 to 60 across samples) is much lower than the number of reads analysed by HMMSTR (ranging from 400 to 1000 for the same set of samples).
So, the question would be: are those SEQS all "unique" sequences (namely all haplotype variants) or if any of those haplotype variants is supported by multiple reads, then it would be reported only once?
Basically, I am asking if processing that output I can obtain the distribution of repeat lengths inferred by the software or just all possible haplotypes variants.
Thanks,
SM

[wdecoster]

Right, I understand. No, STRdust takes a sample of the reads (for computational efficiency), but you can change the parameters to get more

[MaestSi]

Ok, thanks. Is --consensus-reads the parameter I should tune, as the default value is 20? Or are there any other parameters for analysing more reads I should be aware of?
SM

[wdecoster]

I am updating the documentation now, you need --max-number-reads -1 to use all reads.