aidenlab / juicer

A One-Click System for Analyzing Loop-Resolution Hi-C Experiments
http://aidenlab.org
MIT License
402 stars 180 forks source link

Missing merged loops HiCCUPs output #296

Open ericmalekos opened 1 year ago

ericmalekos commented 1 year ago

Hello, I used hiccups --cpu to call loops and it completed without error producing all of the expected files except the merged_loops list. I ran with

java -Djava.awt.headless=true -jar juicer_tools.2.20.00.jar hiccups --cpu -r 1000 -k KR --threads 16 norm_Ctrl.hic HiccupOut

And had the following output,

WARN [2022-11-01T07:35:34,054]  [Globals.java:138] [main]  Development mode is enabled
Reading file: norm_THP1.Ctrl.hic
Using the following configurations for HiCCUPS:
Config res: 1000 peak: 5 window: 10 fdr: 10% radius: 3000
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 16 CPU thread(s) for primary task  
Unable to assess map sparsity; continuing with HiCCUPS
Running HiCCUPS for resolution 1000
Data not available for 9 at 1000 resolution
Data not available for 12 at 1000 resolution
Data not available for 15 at 1000 resolution
Data not available for 18 at 1000 resolution
Data not available for Y at 1000 resolution
4%
9%
...
94%
99%
154690 features written to file: /HiccupOut/enriched_pixels_1000.bedpe
14682 features written to file: /HiccupOut/postprocessed_pixels_1000.bedpe
HiCCUPS complete

I'm not sure how the pixel.bedpe files relate to the merge_loop list... Can I infer the loops from them? Thanks!

ericmalekos commented 1 year ago

It's possibly related to the resolution of 1kb. I reran on a different .hic at 5kb and it produced the expected output files.