Hello everyone,
I have been trying to perform GISTIC analysis in Linux on methylation CNV data of ~650 tumor samples.
I use the following command:
gistic2 -b <directory> -seg gistic.segments -refgene hg19.mat conf -0.95
The head of the gistic.segments file is as follows: ID chrom loc.start loc.end num.mark seg.mean
Sample1 chr1 560684 145225000 862 0.156
Sample1 chr1 145375000 249195311 644 -0.13
Sample1 chr10 105000 107175000 589 -0.02
Sample1 chr10 108425000 135462374 202 -0.273
Sample1 chr11 130000 134873258 881 -0.149
Sample1 chr12 172870 34375000 207 0.121
Sample1 chr12 34425000 34703347 4 -0.424
Sample1 chr12 38178347 133770948 607 -0.107
Sample1 chr13 19110000 115079939 416 -0.046
The tool works perfectly if I run ~430 samples or less. In this case it doesn't matter which samples I select out of the 650. However, if I try to run more samples I run in to an error due to a floating point exception:
Finding peaks on chromosome 1
Finding peaks on chromosome 2
Amplification peak at chr2:204700000-206675000(36452:36650)
q-value = 0.0021198 & residual q-value = 0.0021198
Finding peaks on chromosome 3
Amplification peak at chr3:8500000-9525000(39381:39467)
q-value = 0.005833 & residual q-value = 0.005833
Finding peaks on chromosome 4
Resolving conflicts...
Amplification peak at chr4:54625000-54975000(54479:54514)
q-value = 1.023e-22 & residual q-value = 1.023e-22
Finding peaks on chromosome 5
Finding peaks on chromosome 6
Floating point exception detected at Wed Jan 15 13:52:45 2020
Configuration:
Crash Decoding : Disabled
Current Visual : None
Default Encoding : UTF-8
GNU C Library : 2.27 stable
MATLAB Architecture: glnxa64
MATLAB Root : /usr/share/mcr/v83
MATLAB Version : 8.3.0.532 (R2014a)
Operating System : Linux 4.15.0-51-generic #55-Ubuntu SMP Wed May 15 14:27:21 UTC 2019 x86_64
Processor ID : x86 Family 6 Model 62 Stepping 4, GenuineIntel
Virtual Machine : Java 1.7.0_11-b21 with Oracle Corporation Java HotSpot(TM) 64-Bit Server VM mixed mode
Window System : No active display
Fault Count: 1
Abnormal termination:
Floating point exception
The chromosome at which the tool crashes varies (in this case chromosome 6), but it will crash somewhere if I use more than 430 samples.
Does anybody know what I'm doing incorrectly?
Thank you very much for your help!
It is not an answer at all: what is methylation cnv data? How did you get it?
In my experience all the errors from gistic were caused by faulty samples (one per cohort with one "bad" cnv) or wrong markers file. Can you try to support your run with manually generated markers file, taking into account that all of your cnvs have to contain at least one marker within?
I merged conumee segments files based on 450k data (which uses the DNAcopy package for this analysis).
The merged segments file works perfectly if I randomly select ~430 samples out of the 650. However, problems appear if I try to add more samples. I haven't been able to point out one bad sample, because all samples work individually if I keep my total of samples low enough.
I have run gistic2 without a markers file since it wasn't obligatory, but I'll try to create one and see if it works.
Thank you very much for your suggestions!
I merged conumee segments files based on 450k data (which uses the DNAcopy package for this analysis).
The merged segments file works perfectly if I randomly select ~430 samples out of the 650. However, problems appear if I try to add more samples. I haven't been able to point out one bad sample, because all samples work individually if I keep my total of samples low enough.
I have run gistic2 without a markers file since it wasn't obligatory, but I'll try to create one and see if it works.
Thank you very much for your suggestions!