VNTR motif count

[tzhu-bio]

Thanks for the great tools. I've noticed that at many VNTR sites, the copy number estimated by TRGT seems unreasonable. Could it be that in ALT, these might not even qualify as VNTRs anymore? For example:

Ref: ATATATATATATAATATAATTATAATC count number is about 2 (reasonable).

The TRGT output the ALT count number is about 134.

Based on the reads, can this region even be considered a VNTR? If it's not a VNTR, does calculating its copy number become meaningless?

[pbsena]

Hello,

Thank you for reporting this example. Would you also be able to share (a) the command call (usually on the header of the output VCF in the ##trgtCommand line) and (b) what the VCF line for this example looks like? That may help us understand the reports a bit better.

Just from the image of the longer allele, my intuition is that you're right that this region is either not repetitive, or that there are other motif units here that weren't quantified by target (likely not part of the input BED).

Thank you!

[tzhu-bio]

Thanks for your reply.

Here this the command:
##trgtVersion=4.0.0-429afcd
##trgtCommand=/work/home/acc1hz5x99/zhutao/software/trgt-v4.0.0-x86_64-unknown-linux-gnu/trgt genotype --genome /work/home/acc1hz5x99/00_longfei/01_graph_based_genome/W_genome_assembly.fasta --repeats W_trf_tr.bed --reads /work/home/acc1hz5x99/zhutao/pbmm2/CS001.aligned.bam --output-prefix CS001 -t 10 --sample-name CS001
##trgtCommand=/work/home/acc1hz5x99/zhutao/software/trgt-v4.0.0-x86_64-unknown-linux-gnu/trgt merge --genome /work/home/acc1hz5x99/00_longfei/01_graph_based_genome/W_genome_assembly.fasta --vcf-list vcf_file_list_sorted.txt --output merged_vcf

Here is the VCF line for this example:

6_16132342.vcf.txt

[tzhu-bio]

Could you help me resolve this issue?

[egor-dolzhenko]

Hi @tzhu-bio. I agree -- some repeat regions are so degenerate that it might not make sense to consider them as VNTRs. However, introducing a hard threshold that would separate VNTRs from non-VNTRs is tricky and can introduce artifacts in many types of repeat analyses.

TRGT handles this issue like this: If an allele sequence (or a part of it) has higher similarity to a random sequence than to a repeat composed of the specified motifs, it will be left unsegmented (and will be displayed as a gray bar in TRGT plots). In your example, TRGT decided that the 4,745bp allele has closer resemblance to a repeat with ATATATATATATAATATAATTATAATC motif than to a random sequence and hence proceeded with annotating it and counting motifs.

So TRGT's ability to handle very degenerate VNTRs like the one in your example is by design. It is useful for some analyses and might not be useful for others. If you'd like to focus your analysis on relatively pure VNTRs, you can filter alleles based on repeat purity (reported in the AP filed).

Does this sound reasonable? Please let me know if you have any additional thoughts about this.