Prediction of deletion of a base in repeat region

Question

[Amk]

Dear All,

We have done Variant calling using Illumina Basespace suite. The vcf file which have been generated was annotated using Variant studio. In the report we came to see a deletion of TA > TA/A . As that region when we viewed in the IGV it shows the picture below(Link).

https://drive.google.com/file/d/0B6z33yOKO0MhWTJtNVI5VXhFR00/view?usp=sharing

As in that picture after the T, it is a long repeats of A. The deletions are shown right adjacent A of T. How can we be sure that the deletion happened in that A itself? We think deletion can occur on any other A in that stretch, as the deletion in the report shows the change of TA > TA /A. How the tool predicts it is the adjacent A itself?

Thank you in advance.

Amk

[Jenez]

Out of curiosity, why would it matter which of the bases was deleted? The end result will be the same regardless of which one it was deleted, or am I missing something exciting?

[Amk]

Hi,

Actually that site belongs to a splice region. As you know the variations in that regions also causes various problems. Here our concern is about the tools on prediction of the deletion on the repeated region. Thank you.

Amk

[Devon Ryan]

Since the splice site isn't affected by the deletion (it doesn't matter which A is deleted, the splice acceptor/donor site has the same sequence) you can ignore any SNPeff or similar predictions of altered splicing.

Answer 1

[Devon Ryan]

As you surmised, you can't actually know which of the A's are deleted. When variants are called, things are either right or left justified in the process.

[Amk]

Hi Devon,

As this site is on starting of a splice junction, it is a valuable information that which is the exact position of deletion occurred. If the variant caller justified the reads to one side and find the deletion, how can we say it is the first A got deleted?, it can be the last one or the middle one. As here how can it identify its the A near the T itself. Will it be a biased result if repeats like this come into play?
Thank you for your time.
Amk

[Devon Ryan]

It doesn't matter which one is deleted, the splice site isn't affected. It has the same sequence regardless of which A got deleted.

[Amk]

Hi Devon,

Apart from biology, As a curiosity I like to know how these variant callers deal with the situations like this. My concern is to know if this is cause a bias in the callers, to point the base correctly which have been deleted, when we have a region of interest which can play important role in diseases due to these mutations.

Thank you.
Amk

[Devon Ryan]

You might be interested in bcftools norm, which will normalize VCF files so you get the same representation regardless of variant caller. For how each caller deals with situations like this, you'll have to read its documentation. This is sometimes, but not always, mentioned.