gistic2.0.23 error:Index exceeds matrix dimensions. Error in normalize_by_arm_length

[yiqing zhang]

Hello,

I am trying to run GISTIC on a segments file but can't seem to solve an error.

GISTIC version: v.2.0.23

Input file: segments file from WES, with columns sample, chrom , chr_start, chr_stop, num_markers and seg_mean. I create this file using varscan. Simultaneously input the self created marker file.
Genome version: hg38

The error is:

Index exceeds matrix dimensions.
Error in normalize_by_arm_length (line 85)
Error in make_sample_B (line 50)
Error in perform_deconstruction (line 68)
Error in perform_ziggurat_deconstruction (line 126)
Error in run_focal_gistic (line 151)
Error in run_gistic20 (line 124)
Error in run_gistic2_from_seg (line 249)
Error in gp_gistic2_from_seg (line 97)
MATLAB:badsubscript

Hope you can help us!

Best,

zhangyiiqng0706.

[Andrew Cherniack]

Hi Yiqing,

The issue is likely with your marker file. Try running GISTIC without a marker file and see if it works.

If you create your own marker file there needs to be a marker at every single segment breaks or it will error out.

Also how many samples are in your seg file?

Regards,

Andrew

[yiqing zhang]

Hi Andrew ,

I have tried to run without using the marker file, but he still reported this error. Additionally, I have 18 samples. This is my seg file. Can you help me troubleshoot the error?

[Andrew Cherniack]

Hi Yiqing,

Your seg file is badly hypersegmented which is causing GISTIC to error out.

Seg files for each tumor  usually have a few 100 segments or less.

You will need trouble shoot the what ever you used to generate CN from WES.

Regards

Andrew

--
You received this message because you are subscribed to a topic in the Google Groups "GenePattern Help Forum" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/genepattern-help/gL1tR0TEYX0/unsubscribe.
To unsubscribe from this group and all its topics, send an email to genepattern-he...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/genepattern-help/7e9d8d0c-c031-48a3-9060-07b0d713d22fn%40googlegroups.com.

[yiqing zhang]

Hi Andrew,

I use the same method but it can run normally when the number of samples is reduced. Can you tell me why?

[Andrew Cherniack]

The issue has less to do with the number of samples than with your Copy Number data itself.

Here are the number of segments that you have in each sample.

tumor	number of segments
X11A	72915
X12A	81039
X19A	102086
X21A	9517
X22A	105243
X25A	20154
X26A	38608
X27A	7407
X31A	29481
X38A	46123
X39A	130194
X40A	271113
X44A	13988
X45A	96492
X50A	142648
X52A	153809
X54A	161270
X56A	164723

 Almost all of your tumors have 10s to 100s of thousands of copy number segments.

This means your data is extremely noisy and something went wrong either with your sequencing or the algorithm you used to generate your copy number data.

The total number of CN segments for tumors should be in the 100s and no more than a few 1000.  GISTIC cannot handle data with this much noise and so it is errroring out.

Because of this, GISTIC's default setting is set to ignore tumors with more than 2500 segments.

I am sorry, but I cannot help you with troubleshooting your sequencing or the pipeline that you used to generate your CN.

Regards,

Andrew

You received this message because you are subscribed to the Google Groups "GenePattern Help Forum" group.
To unsubscribe from this group and stop receiving emails from it, send an email to genepattern-he...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/genepattern-help/180346e4-7192-4b4b-a3de-17b1ddca29bdn%40googlegroups.com.

[Andrew Cherniack]

he issue has less to do with the number of samples than with your Copy Number data itself.

Here are the number of segments that you have in each sample.

X11A:  72915

X12A:  81039
X19A: 102086
X21A:  9517
X22A: 105243
X25A: 20154
X26A: 38608
X27A: 7407
X31A: 29481
X38A: 46123
X39A: 130194
X40A: 271113
X44A: 13988
X45A: 96492
X50A: 142648
X52A: 153809
X54A:161270
X56A: 164723

 Almost all your tumors have 10s to 100s of thousands of copy number segments.

This means your data is extremely noisy and something went wrong either with your sequencing or the algorithm you used to generate your copy number data.

The total number of CN segments for tumors should be in the 100s and no more than a few 1000.  GISTIC cannot handle data with this much noise and so it is errroring out.

Because of this, GISTIC's default setting is set to ignore tumors with more than 2500 segments.

I am sorry, but I cannot help you with troubleshooting your sequencing or the pipeline that you used to generate your CN.

Regards,

Andrew

[Andrew Cherniack]

One more thing.  One more indication that there is something wrong with CN data is that you have segment breaks in the middle of intragenic regions that your WES data does not cover.

(see example below)

Are you trying to use off-target reads to generate copy number?  If so this could be messing you up.