Variant calling

Question

[Sabeen]

Dear readers,

Every one has been very helpful in answering my questions. Here i have one more.

I would like to know from detection of a variant in the vcf file , what are the steps to finally classify it as pathogenic or variant of unknown significance.(VUS).

What i am doing so far is filtering the genes based on read depth (>10), those that fit in heterozygous or homozygous category, gnomad frequency <1%., and phenotype match. After getting the short listed variants I will match the variant location on chromosome with the specific NM_ no. (transcript number). After that i will search that NM_ number in the literature to search if it is associated with disease or not. As i believe there are steps that I am missing like type of mutation, synonymous or non synonymous mutation etc Can some please tell me how to go about these steps after variant detection in vcf file. Also which databases I could use to get this extra information that i am missing.

Thanks. sabeen

[nihilior]

Supposing that you are talking about human data, I would also take into consideration prediction scores like CADD or clinical interpretation tools such as Intervar. By "I am missing steps like type of mutation, synonymous or non synonymous mutation" you mean that you do not have such annotations? In that case, you should include this kind of annotation to your vcf files with softwares like VEP, Annovar or SnpEff.

[Sabeen]

Thank you for your reply was very useful esp using CADD. The ACMG guidelines also recommend that. But I am having CADD score in my annotated file however according to ACMG I need more information like effect on amino acid change , deleterious effect , if the deleterious effect is only when truncating gene etc. Can you guide me how can I get that information .

[nihilior]

You have to annotate your vcf file with a software like VEP, using the proper caches (https://www.ensembl.org/info/docs/tools/vep/script/vep_cache.html) . Then, you can add some VEP plugins to your annotation, even custom ones, such as Clinvar for "deleteriousness".

Answer 1

[Prash]

Sabeen Plan A: If you are a naive programmer, you could use snp-nexus.org and upload your fiiltered VCF files so as to get the list of variants across all populated database You will get per variant/text files annotated from all well known databases

Plan B, which I always suggest: Pl use bcftools filter option with a minimum depth, followed by het filters and then use varsome

You could also use varsome or download clinvar tx file from NCBI which has the list of latest variants updated. You coudl awk/sed liners to map your final list of variants

Hope this helps Prash