HICCUPs reported Exception in thread "pool-2-thread-1" java.lang.ArrayIndexOutOfBoundsException: 5980

150 views
Skip to first unread message

曹唱唱

unread,
May 26, 2020, 2:23:11 AM5/26/20
to 3D Genomics
Dear all, 

I generated a HiC dataset for a very small genome (length: < 30kb) and want to call loops from this data. 


My command is: 

java -jar ./software/juicer_tools_1.21.01.jar hiccups --cpu -m 1000 -c chr1 -r 5 -k KR -f 0.1 -p 4 -i 10 -t 0.02,1.5,1.75,2.0 -d 30 --threads 1 --ignore-sparsity small.genome.hic ./loops_call

The running log reported:

Reading file: small.genome.hic 
Using the following configurations for HiCCUPS:
Config res: 5 peak: 4 window: 10 fdr: 10% radius: 30
WARNING - You are using the CPU version of HiCCUPS.
The GPU version of HiCCUPS is the official version and has been tested extensively.
The CPU version only searches for loops within 8MB (by default) of the diagonal and is still experimental.
Using 1 CPU thread(s)
Unable to assess map sparsity; continuing with HiCCUPS
Running HiCCUPS for resolution 5
3%
7%
10%
14%
17%
21%
Exception in thread "pool-2-thread-1" java.lang.ArrayIndexOutOfBoundsException: 5980
        at juicebox.tools.utils.juicer.hiccups.GPUController.runCPUVersion(GPUController.java:384)
        at juicebox.tools.utils.juicer.hiccups.GPUController.process(GPUController.java:141)
        at juicebox.tools.clt.juicer.HiCCUPS.runCoreCodeForHiCCUPS(HiCCUPS.java:521)
        at juicebox.tools.clt.juicer.HiCCUPS.access$200(HiCCUPS.java:158)
        at juicebox.tools.clt.juicer.HiCCUPS$1.run(HiCCUPS.java:434)
        at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
        at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
        at java.lang.Thread.run(Thread.java:748)


3%
7%
10%
14%
17%
21%

It seems that the number 5980 is related to the size of the genome (resolution 5 * 5980 is equal to the length of the genome).

Any suggestions will be appreciated.

Thanks very much.


曹唱唱

unread,
May 26, 2020, 5:08:24 AM5/26/20
to 3D Genomics
It seems that "there is really some issue with non-standard resolutions (either 1kb or 2.5kb, or even 5 bp)"

Solution: All the coordinate is multiplied by a factor 1000, and re-run the pre and hiccups command in which the resolution was set as 5000. Correspondingly, the coordinate for the loops called by hiccups should be divided by a factor 1000.

Reference:
https://github.com/aidenlab/Juicebox/issues/898

在 2020年5月26日星期二 UTC+8下午2:23:11,曹唱唱写道:

HELEN HUANG

unread,
May 19, 2021, 9:24:21 PM5/19/21
to 3D Genomics
Does anyone know if this issue with non-standard resolution has been solved?
Or do we still have to use this coordinate manipulation hack?
Reply all
Reply to author
Forward
0 new messages