exome copy number and LOH

[rcorbett]

Hi folks,
Can anyone recommend a good tool for exome copy number analysis on Illumina data?

Ideally, but not necessarily, it would be a tool that includes GC correction.

We have an in-house tool that was built up on Genome data, but applying the same tool to exome data is tricky due to the higher noise, and sporadic coverage

[gprakhar]

Hello,

Use CNVnator from the 1000 Genomes guys.
But you will have to be careful at the exome boundaries, its best to filter out such CNV calls.

Regards,
Prakhar Gaur

[ulz_peter]

Hi,

I don't see any statement that CNVnator works for exome data. I imagine that is hard as the target enrichment process would introduce a heavy bias in read count per exome...

Any counter-arguments welcome...

[gprakhar]

Hello,

True CNVnator does not work for Exome data. That is why I said
"be careful at the exome boundaries, its best to filter out such CNV calls."

This can be done using a script which has the co-ordinates for the regions enriched and hence can filter out all CNV at the boundaries and the Intron regions. Any how the interest is only in the CNV regions inside Exons so while that can be detected, the CNV breakpoint detection might not be possible if it lies in an Intronic region.

To best of my knowledge there is no tools for CNV detection that can work with out any manual intervention with Exonic data.

Regards,
Prakhar Gaur

[rcorbett]

Thanks for the helpful responses. I'm a little surprised that there aren't any tools that aren't tailored for exome CNV. However, I haven't been able to find anything via google, etc.

Anyone else have a suggestion?

[gprakhar]

What kind of data do you have?
As in, is it a paired Cancer Normal data?

Because if it was then you can use a paired approach which minimizes the False Positive rate due to enrichment. As the introns are common to both and hence are neglected.

Regards,
Prakhar Gaur

[NextGenSeq]

CLC Bio has a copy number variation tool that works with exome data. I personally am skeptical that exome hybridization is quantitative however.

[rcorbett]

Thanks everyone.
The samples I usually look at are tumour normal pairs of exome data. The best results I have gotten are achieved by filtering reads that align outside baits. This as mentioned above reduces bias created from varying bait efficiency. Then I use an HMM to partition into Copy Number regions.

I have samples where we get reasonable looking CNV results, albeit at a much lower resolution that with genomes. I am just checking with everyone to see which, if any, tools have been advertised specific to exome data.

[gprakhar]

Dear rcorbett,

CNAseg http://www.ncbi.nlm.nih.gov/pubmed/20966003, the approach is bit similar to yours.

For a paired data you can also use CNVseq http://tiger.dbs.nus.edu.sg/CNV-seq/ , works ok with paired data and that exome only problem should not trouble you much with a paired data set.
Though you might need to do some filtering at the end.

Has anyone used the CLC Bio CNV caller that works with exome data??

Regards,
pg

[gprakhar]

Hello,

This might be of interest, http://www.biomedcentral.com/1471-2164/12/184

Regards,
pg

[NGSfan]

thanks gprakhar for the link

here is another recently released exome CNV program:

http://www.ncbi.nlm.nih.gov/pubmed?term=21828086

I have not tried it yet

[B.Abrahams (Einstein)]

ExomeCNV great

Hi All,

We have implemented this locally and had great success. Very solid results so long as the intersample correlations meet minimum thresholds. Hope this is helpful.

Brett

[NGSfan]

Hi Brett,

Are you talking about ExomeCNV? or the other one here http://www.biomedcentral.com/1471-2164/12/184 ?

[B.Abrahams (Einstein)]

ExomeCNV from stan nelson's group at UCLA

[Mamun]

0 coverage in the exon parsed coverage files

ExomeCNV from stan nelson's group at UCLA

Hi,
I have been trying to implement ExomeCNV for analysing some
copy number variation data. It requires some data pre-processing.

I have downloaded bunch of perl and shell script from their ftp and
used them for processing

BAM Files > pileup > wig > exon_parsed_wig > exon parsed coverage file.

but apparently the coverage files comes with 0 in average coverage column.

I would really appreciate if anybody can shed any light on it ..


thanks in advance.
Mamun