CONTRA for CNV detection. troubleshooting

Question

[Kizuna]

Hi Biostars Users,

I would like to know if you have encountered this issue while working with CONTRA tool for detecting CNVs in targeted NGS.

In fact, my analysis is "stopping" at the binning process with no error message, I have checked all the required arguments and they seem to be ok :

contra apple$ python contra.py --target ngs1_probes.bed --test cic04668.bam --control  baseline_ngs1.bed --bed --fasta human_ref.fasta --outFolder ~/contrases

target              : ngs1_probes.bed
test                  : cic04668.bam
control                        : baseline_ngs1.bed
fasta                : human_ref.fasta
outfolder        : /Users/apple/contrases
numBin                       : [20]
minreaddepth            : 10
minNBases     : 10
sam                 : False
pval                 : 0.05
sampleName  : No-SampleName
nomultimapped         : False
plot                 : False
bedInput                    : True
minExon                     : 2000
largeDeletion : False
Creating Output Folder :  Done.
Converting TEST Sample...
DEBUG 123 genomeCoverageBed -ibam cic04668.bam -bga -g /Users/apple/contrases/buf/sample.Genome
Converting CONTROL Sample...
Getting targeted regions DOC...
chr1
chr10
chr11
chr12
chr14
chr15
chr16
chr17
chr19
chr2
chr20
chr3
chr4
chr5
chr6
chr7
chr8
chr9
chrX
Targeted regions pre-processing: Done
Getting targeted regions DOC...
chr1
chr10
chr11
chr12
chr14
chr15
chr16
chr17
chr19
chr2
chr20
chr3
chr4
chr5
chr6
chr7
chr8
chr9
chrX
Targeted regions pre-processing: Done
Test file read depth =  97323185
Control file read depth         =  98289477
Pre-processing Completed.
Getting the Log Ratio
Binning....

Do you have any idea why?

Thank you for your help

Kiz

[lffu_0032]

do not anyone think that the CONTRA's paper published in bioinfomatics is confusing? i read the paper but i think there are much logistic problem in the presentation or calculation.

Answer 1

[Kizuna]

Hi sruthimouli88,

To be honest, I just dropped using CONTRA and I am currently using ExomeDepth from (R),

I have configured it and it is working really well.. I advise you to use it, and if you need the scripts for CNVs detection, just tell me :)

I hope this helps,

Kizuna

[sruthimouli88]

@Kizuna: I found out what the issue was in my case. Sharing it here just in case someone refers to this post. The issue with unlimited binning in my case was with --minExons.

--minExon

Minimum number of exons in one bin (if less than this number, bin that contains small number of exons will be merged to the adjacent bins) [Default : 2000]

There were only 800 exons in my targeted region. As this was less than defat value(2000) the tool kept looking for more 1600 exons to include in the bin. As they were not available it just kept binning for a really long time. I changed the option to --minExon=100 (running the 800 exons in 8 bins) and it worked fine.

[Noa]

This was the problem in my case too, thank you for sharing this! very helpful!!

[amitgsir]

Great~~worked for me. But changing thr --minExom provides the diffrent values in *.CNATable.10rd.10bases.20bins.txt file. Is it okay?

Answer 2

[sruthimouli88]

Hello Kizauna,

I am new to CONTRA and I am facing the same problem. I have no idea what is wrong as there is no error message. I am sure the thread is not stopped but it is running in my server for a couple of days. Were you able to resolve this.

Thanks in advance

Answer 3

[Ettore Z]

It could be a Python Memory Error issue. Try redirecting the standard error log in a file, i.e.

python contra.py \
  --target ngs1_probes.bed \
  --test cic04668.bam \
  --control baseline_ngs1.bed \
  --bed \
  --fasta human_ref.fasta \
  --outFolder ~/contrases 2> error.log #like this, with the 2> redirect

Answer 4

[Avinash]

Hi Kizuna,

I am new to CNV analysis and beginner in R language. I am trying to call germline CNVs using exome data using ExomeDepth. I only have the raw data with hg38 reference.

If you have the ExomeDepth scripts to run on hg38 reference. Kindly share me the scripts. It would be a great help for me to continue the analysis. Also share me the hg19 current script that you are without any bugs.

I have tried CNVkit but it was less accurate in detecting CNVs smaller than 1 Mbp but my samples are mostly less than 1Mbp duplication or deletion. So I tried to shift with other tools ExomeDepth, Conifer, GATK gCNV, CONTRA etc. If you have any successful tools that works well fits for my scenario. Kindly suggest me.

Awaiting for your response.

Thanks in advance.

Avinash