Find DNA deletion with illumina 100bp pair-end reads

[remimaglione]

Hi,

I'm quite new in NGS. I'm looking for tools/stuff/idea to find DNA deletion. I'm working with illumina 100bp pair-end reads.
I have try to find lots of kind of tools/stuff on this forum to deal with DNA deletion.

I've tried using the classic pipeline Bowtie2/topHat2, but there are some ideas that disturbs me with this.
Indeed, tophat will seek to identify (donor sites / receiver) canonical splicing signals.
This is the strength of tophat for RNA-seq, but it does not interest me in my study.
On the other hand, Tophat gives me a outpout with an empty file deletion.bed for my candidate sample and full of things for my control sample...

In addition, I tried Pindel software, which is design to find deletion / indel / junction / etc, but again, this software is very limited for my analysis.
In fact, it only considers the reads themselves and not the size of the insert between the two mate of the same pair.

Finally, I used subread, but it manages only 16bp indel of maximum and DNA degradation/deletionI am looking for is quite large in size.

Currently, I am trying to perform an analysis with subjunc, which was developed to find the junctions, which in a sense could help me in my search for potential deletion.
Indeed, the first step to find the junction seems powerful, but the validation stage based on control of canonical splicing signals could be problematic.

So I look to the community if anyone can help me in my search for deletion.

thank you in advance

[cwhelan]

How big is the deletion you are looking for? Do you have RNA-seq data or whole genome sequencing?

If you know exactly where it is you might try simply looking for discordant read pairs that map to either side of the deletion. Visualize your BAM files in IGV as pairs and look at your region of interest.

If you have DNA sequencing data and want to use a program to look for deletions based on insert sizes I'd recommend either Breakdancer or DELLY.

[remimaglione]

How big is the deletion you are looking for?[...]If you know exactly where it is[...]

I really do'nt know. I'm trying to evaluate DNA degradation by 1rst looking for DNA deletion.

Do you have RNA-seq data or whole genome sequencing?

i'm working on whole genome sequencing. Sorry, I forgot to mention it.

Visualize your BAM files in IGV as pairs and look at your region of interest.

Already done with my bam come from BWA, bowtie2 amd topHat2. But it's quite hard because my genome is around 100Kpb with 1000 to 3000X mean coverage of reads (with peak at 8000X)

If you have DNA sequencing data and want to use a program to look for deletions based on insert sizes I'd recommend either Breakdancer or DELLY.

This can be really interesting for what I want. I will study these programs more closely.

thank you for your quick response.

Any idea why my tophat outpout (deletion.bed) is empty for my candidate sample ?

[shi]

Hi,

I'm quite new in NGS. I'm looking for tools/stuff/idea to find DNA deletion. I'm working with illumina 100bp pair-end reads.
I have try to find lots of kind of tools/stuff on this forum to deal with DNA deletion.

I've tried using the classic pipeline Bowtie2/topHat2, but there are some ideas that disturbs me with this.
Indeed, tophat will seek to identify (donor sites / receiver) canonical splicing signals.
This is the strength of tophat for RNA-seq, but it does not interest me in my study.
On the other hand, Tophat gives me a outpout with an empty file deletion.bed for my candidate sample and full of things for my control sample...

In addition, I tried Pindel software, which is design to find deletion / indel / junction / etc, but again, this software is very limited for my analysis.
In fact, it only considers the reads themselves and not the size of the insert between the two mate of the same pair.

Finally, I used subread, but it manages only 16bp indel of maximum and DNA degradation/deletionI am looking for is quite large in size.

Currently, I am trying to perform an analysis with subjunc, which was developed to find the junctions, which in a sense could help me in my search for potential deletion.
Indeed, the first step to find the junction seems powerful, but the validation stage based on control of canonical splicing signals could be problematic.

So I look to the community if anyone can help me in my search for deletion.

thank you in advance

Dear remimaglione,

You are correct that the current version of Subread can only detect up to 16bp indels. But we have an in-house 1.4 version of it which might help to detect long indels.

Have a look my previous post on how to access this version and the command for detecting long indels.

http://seqanswers.com/forums/showpost.php?p=104220&postcount=6

Hope this helps.

Best wishes,

Wei

[KaiYe]

Pindel uses both read-pair and split-read for variant detection. just an update.