General question about usage

[HLHsieh]

Hi Mark and Bida,

Thank you for your prompt response regarding the generation of motif sets. I opened a new issue to ask additional questions about the usage.

1. According to the help manual, vamos includes four subcommands:

vamos --contig [-b in.bam] [-r vntrs_region_motifs.bed] [-o output.vcf] [-s sample_name] [-t threads]
vamos --read [-b in.bam] [-r vntrs_region_motifs.bed] [-o output.vcf] [-s sample_name] [-t threads] [-p phase_flank]
vamos --somatic [-b in.bam] [-r vntrs_region_motifs.bed] [-o output.vcf] [-s sample_name] [-t threads] [-p phase_flank]
vamos -m [verison of efficient motif set]

So far, only contig and read have been introduced in the documentation. I would like to ask whether somatic is intended for detecting somatic instability. I would greatly appreciate it if you could provide more details on the purpose and use cases of somatic and m subcommands.

2. Does vamos provide read-level information corresponding to each haplotype, as Mark mentioned that the tool partitions reads prior to annotating tandem repeats? If so, is it possible to examine the supporting reads for each allele? This would help in refining the motif set and improving the annotation by rerunning with additional or more appropriate motifs.

3. I applied vamos to several mock samples, but the genotyping results did not match the expected values.

For example, in a mock sample with 16 RU, I executed the following command: vamos --read -b C9ORF72_1.sorted.bam -r C9ORF72.tsv -s C9ORF72_1 -o C9ORF72_1.vcf

The corresponding C9ORF72.tsv file was:

chr9	27573528	27573546	GGCCCC

However, the VCF output was:

#CHROM	POS	ID	REF	ALT	QUAL	FILTER	INFO	FORMAT	C9ORF72_1
chr9	27573528	.	N	<VNTR>	.	PASS	END=27573546;RU=GGCCCC;SVTYPE=VNTR;ALTANNO_H1=0-0-0-0-0-0;LEN_H1=6;	GT	1/1

In this case, the reported length (LEN_H1=6) is inconsistent with the expected repeat count of 16 units.
I am wondering if this discrepancy could be due to an issue with my input or parameter settings. Could you please advise whether there is anything I might need to modify or check?

4. In the VCF output, I noticed that the ID column always shows a dot (.). Is there any way to customize or assign a specific identifier to each VNTR record in the output?

5. As mentioned earlier, the tool is currently limited to diploid genotyping. Does this imply that mosaicism detection is also not supported under the current framework? If so, are there any recommended strategies or alternative approaches for detecting or visualizing mosaic repeat structures using vamos?

I would greatly appreciate any insights or suggestions regarding the questions above.

Best,
Hsin

[mchaisso]

Hi Hsin,
We have rudimentary support for somatic variation using the --somatic flag. The output is a bed file (I just pushed a commit to master that fixes output.vcf to output.bed. This produces a bed file with the coordinates of the tandem repeat in GRCh38 (or CHM13T2T if that is the reference you are using) followed by the motifs used to annotate that locus, followed by a column that reflects the annotations of reads from each haplotype. The annotation has:

hap_i:size_i:a_^i_1,a^i_2,...,a^i_size_i

for all reads i = 1...m overlapping a locus, where hap_i is the haplotype (phase) of read i, size_i is the number of repeated motifs, then a^i_1, ... a^i_size_i is the list of annotation indexes for the read.

I haven't fully tested this yet since the data I have is not really showing somatic variation. If you are able to share any data, I can make sure it works.

The -m flag gives a list of commands you can use to download the input repeat tables.

For (2) the --somatic flag does this.

For (3) - are you able to share a read? I can debug the example if so.

For (4) - I can update the output to reference the corresponding entry from the input table.

For (5) - If it is mosaicism in a diploid sample, where only the tandem repeat is mosaic, the --somatic should work for this. The phasing algorithm assumes diploid SNVs outside the tandem repeat though, so the read partitioning will be incorrect in a polyploid genome.

Happy to chat directly if you have additional questions or data you can share.

Best,
-Mark

Hi Mark and Bida,

Thank you for your prompt response regarding the generation of motif sets. I opened a new issue to ask additional questions about the usage.

1. According to the help manual, vamos includes four subcommands:

vamos --contig [-b in.bam] [-r vntrs_region_motifs.bed] [-o output.vcf] [-s sample_name] [-t threads]
vamos --read [-b in.bam] [-r vntrs_region_motifs.bed] [-o output.vcf] [-s sample_name] [-t threads] [-p phase_flank]
vamos --somatic [-b in.bam] [-r vntrs_region_motifs.bed] [-o output.vcf] [-s sample_name] [-t threads] [-p phase_flank]
vamos -m [verison of efficient motif set]

So far, only contig and read have been introduced in the documentation. I would like to ask whether somatic is intended for detecting somatic instability. I would greatly appreciate it if you could provide more details on the purpose and use cases of somatic and m subcommands.

2. Does vamos provide read-level information corresponding to each haplotype, as Mark mentioned that the tool partitions reads prior to annotating tandem repeats? If so, is it possible to examine the supporting reads for each allele? This would help in refining the motif set and improving the annotation by rerunning with additional or more appropriate motifs.
3. I applied vamos to several mock samples, but the genotyping results did not match the expected values.

For example, in a mock sample with 16 RU, I executed the following command: vamos --read -b C9ORF72_1.sorted.bam -r C9ORF72.tsv -s C9ORF72_1 -o C9ORF72_1.vcf

The corresponding C9ORF72.tsv file was:

chr9	27573528	27573546	GGCCCC

However, the VCF output was:

#CHROM	POS	ID	REF	ALT	QUAL	FILTER	INFO	FORMAT	C9ORF72_1
chr9	27573528	.	N	<VNTR>	.	PASS	END=27573546;RU=GGCCCC;SVTYPE=VNTR;ALTANNO_H1=0-0-0-0-0-0;LEN_H1=6;	GT	1/1

In this case, the reported length (LEN_H1=6) is inconsistent with the expected repeat count of 16 units. I am wondering if this discrepancy could be due to an issue with my input or parameter settings. Could you please advise whether there is anything I might need to modify or check?

4. In the VCF output, I noticed that the ID column always shows a dot (.). Is there any way to customize or assign a specific identifier to each VNTR record in the output?
5. As mentioned earlier, the tool is currently limited to diploid genotyping. Does this imply that mosaicism detection is also not supported under the current framework? If so, are there any recommended strategies or alternative approaches for detecting or visualizing mosaic repeat structures using vamos?

I would greatly appreciate any insights or suggestions regarding the questions above.

Best, Hsin

[HLHsieh]

Hi Mark,

Thank you for your kind response.

1. It’s really helpful to provide read-level information. May I ask whether the test.vcf output generated using the --read subcommand is derived from this read-level data? My question stems from the fact that --read and --somatic are separate subcommands, which made me wonder if they operate independently or share the same internal annotation logic.

2. I do not currently have a mosaic sample available but am preparing one. In the meantime, I tested the somatic mode on a mock sample using the following commands:

vamos --read -b C9ORF72_1.sorted.bam -r vamos.bed -s test -o test.vcf -S -t 8
vamos --somatic -b C9ORF72_1.sorted.bam -r vamos.bed -s test -o test.bed -S -t 8

Their output

chr9	27573528	.	N	<VNTR>	.	PASS	END=27573546;RU=GGCCCC;SVTYPE=VNTR;ALTANNO_H1=0-0-0;LEN_H1=3;	GT	1/1:CGGCCCCGGCCCCGGCCC

chr9	27573528	27573546	GGCCCC	0:3:0,0,0;1:3:0,0,0;2:3:0,0,0;1:3:0,0,0;

This sample has 4 reads, each with 3 repeat units, but the somatic output shows three different haplotype groups. I assume this is likely due to the small number of reads.

Besides, could you briefly explain the underlying principle of somatic detection? From what I understand, mosaicism detection relies on variations in repeat length distributions, which seems conceptually similar to what is required for polyploid genotyping. Or perhaps there are technical distinctions I haven’t fully considered—I’d appreciate any thoughts you might have on whether my reasoning is sound, or if there are important limitations I may be overlooking.

Note: This is the same sample I mentioned earlier, which has 16 repeat units. However, after reinstalling vamos, the reported result was 3 units. I’m not sure what caused this discrepancy. I also ran the same sample through TRGT, which returned the expected count of 16. You can access the test dataset here (FASTA and BAM included): https://www.dropbox.com/scl/fo/k4mnzqwowyfhycnvg42at/AIUOdzBHJlET6NulhTEu-CA?rlkey=lmw6h83ekl3clofgqx0fp4bk2&st=swcjm53x&dl=0

Thanks again for your time—I really appreciate your help in understanding these details.

Best regards,
Hsin