Genomic Rearrangement Analysis

[shruti]

Hi,
I am new to NGS analysis and we have a project for identifying genomic rearrangements. I have gone through several papers and in the recent time there has been a rise in the number of different softwares and algorithms. and now I'm confused, how do I choose which one is better. Apart from the insertion, deletion, inversions and translocations I also need to find amplifications (copy number variation). I was thinking of a combination of softwares for the study. What is everyones opinion. Please give me some guidance.

Thanks.
Shruti

[HESmith]

This review is a good starting point.

[Michael.James.Clark]

They're all decent, but none of them does an amazing job. The overlap is decent--greater than 50-70% between them generally. False positives and low sensitivity are both major issues.

The best way to find SVs is to design your experiment appropriately. With current technology, a two-part experiment is best. For human, part one is a low-coverage WGS with 2000bp inserts and part two is medium-coverage WGS with 200bp inserts. The best case is where the insert sizes are a tight distribution centered on these values. This combination is more robust to repeat elements than a typical WGS with a single insert size.

For algorithms, I'm partial to the one I wrote (Breakway), but there are a number of others in common use (see that review for some examples, or check the wiki here).

[shruti]

Thank you all for the feedback.

I had read the reviews and also some of the papers in which most of them make their own algorithms for SV detection.

Yes, it could be a 50-60% of overlap depending upon the kind of signatures they are trying to find. Looks like one cannot use just one tool to identify all types. Though I saw that most people use MAQ for the alignment of reads...

I think you cannot use PEM for study of amplifications. I saw number of algorithms like binary circular segmentation, EWT and SegSeq.. Which one in your opinion would be ideal.