Control-FREEC error message

Question

[2nelly]

Hi all,

I am trying to use Control-FreeC software in order to look for CNA in tumor samples, but I have to deal with an error:

 terminate called after throwing an instance of 'std::bad_alloc'
    what(): std::bad_alloc
Aborted (core dumped)

It seems that program run out of memory. I tried to rerun in single-thread mode but I got the same error. Note that I have plenty of free memory while running freec.

Did anyone face the same problem?

Thank you in advance.

[ivivek_ngs]

Definitely a problem of the memory allocation. Is your config file properly set? All your parameters in the config file set as per the best practice? V. Boeva the developer of the tool is pretty active if you mail her to reply for the problem. There can be many reasons for this problem.

The window and the step parameter setting for the segmentation might be too low, and also theminMappabilityPerWindow and breakPointThreshold you can increase that and see if that is really creating the problem. Then are you running it on whole genome or whole exome? I have usually used it for whole exome with normal and tumor bam files with no segmentation error.

How much memory are you allocating to the script while running the tool on your config file?

[2nelly]

Thank you for your answer Chris. You know I tried to play with all these options you mentioned before. I run a test sample a month ago and it worked like a charm. Now that I decided to run a whole analysis I faced this issue. Since, I was extremely disappointed and I couldn't figure out what is going on, I decided to re-download the freec and guess what.....It WORKED. I dunno, but somehow the previous version I downloaded a month ago was corrupted. The most frustrating thing is that I almost spent a day on that, but at the end the solution was sooooo simple! S...t happens!

In any case, thank you for your answer Cheers

[ivivek_ngs]

Probably a problem with the different archived dependency file paths that might have been screwed or not set earlier with the previous version you downloaded. Was it a different version you downloaded again or the same version?

[2nelly]

The version is exactly the same, so probably the error should be due to some screwed files. Thanks anyway

[longevityamit]

Have you found the solution? I am also facing the same problem: terminate called after throwing an instance of 'std::bad_alloc' what(): std::bad_alloc Aborted (core dumped)

[2nelly]

I just re-downloaded and it worked! strange eh? good luck!

Answer 1

[rse]

What snp bed file did you use?

[2nelly]

I run it without BAF analysis. But, if you are interesting in snp file, you can downloaded from UCSC table browser.

[ivivek_ngs]

I do not think to use a SNP bed file from outside source unless it is formatted that way provided by the tool. If you know that the software comes with its own bundle of packaged set of SNP files and mappability files which one should use.

[rse]

Thank you all. It worked. I used the set of snp files in the CONTROL-FREEC package. I want to get the regions of LOH. But the output shows regions of copy neutral LOH ONLY. What about other types of LOH? Also, the number of CNV Calls were quite high as compared to other tools. Are they false positives?

[2nelly]

False positives!!! Well, this is a good question!I would say yes and no. For instance if you try to play with all arguments included in this package, you will realize that your results have significant differences. For instance different segmentation provides you different results. But this doesn t mean that it s wrong per se. So, I cannot answer this question. Maybe someone more experienced than me can give you a more precise answer

[rse]

Thank you. Yes, I also think so. There must be a way to detect false positives? Like in varscan2, there is a fpfilter perl script provided. Also, i want to detect LOH. Can you suggest any good method?

Answer 2

[rse]

Thank you all. It worked. I used the set of snp files in the CONTROL-FREEC package. I want to get the regions of LOH. But the output shows regions of copy neutral LOH ONLY. What about other types of LOH? Also, the number of CNV Calls were quite high as compared to other tools. Are they false positives?