No loops detected using balanced matrix

[davidecarnevali]

Hi,
I'm having problems detecting loops with hicDetectLoops
When running it using a cool matrix, converted from HicPro raw matrix, many loops are detected. However when running hicDetectLoops on the same cool matrix but balanced with cooler, I got no loops at all.
Here what I did:

First I converted the HiCPro raw matrix to cool format using hicConvertFormat:

hicConvertFormat -m sample1_10kb.matrix --bedFileHicpro sample1.bed --inputFormat hicpro --outputFormat cool -o sample1_10kb.cool

Then I calculated and added the 'weight' column without changing the counts values:

cooler balance sample1_10kb.cool

Running hicDetectLoops on the balanced cool file yelds me no loops at all

Any hint?

Thank you

[joachimwolff]

Cooler does not scale the values back. The KR correction transforms all values to the 0-1 range. In HiCExplorer we expect the data to be in the original data range. If you use our hicCorrectMatrix, we scale the values back. Absolutely no need to use cooler for this. Davide ***@***.***> schrieb am Do. 23. Jan. 2025 um 14:10:

…

Hi, I'm having problems detecting loops with hicDetectLoops When running it using a cool matrix, converted from HicPro raw matrix, many loops are detected. However when running hicDetectLoops on the same cool matrix but balanced with cooler, I got no loops at all. Here what I did: First I converted the HiCPro raw matrix to cool format using hicConvertFormat: hicConvertFormat -m sample1_10kb.matrix --bedFileHicpro sample1.bed --inputFormat hicpro --outputFormat cool -o sample1_10kb.cool Then I calculated and added the 'weight' column without changing the counts values: cooler balance sample1_10kb.cool Running hicDetectLoops on the balanced cool file yelds me no loops at all Any hint? Thank you — Reply to this email directly, view it on GitHub <#921>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADGQCAD7CO2DA2MNBBNRIKL2MDS53AVCNFSM6AAAAABVXKDS76VHI2DSMVQWIX3LMV43ASLTON2WKOZSHAYDMOBXGIZDKNQ> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[davidecarnevali]

Cooler does not scale the values back. The KR correction transforms all
values to the 0-1 range.

In HiCExplorer we expect the data to be in the original data range. If you
use our hicCorrectMatrix, we scale the values back. Absolutely no need to
use cooler for this.

Davide @.***> schrieb am Do. 23. Jan. 2025 um 14:10:
…

Thank you.
So, since hicCorrectMatrix only works on h5 files, I guess I need first to convert raw hicpro matrices to .h5 then run hicCorrectMatrix applying the desired correction *and then convert them to .cool
No need to use firs hicNormalize as suggested right?

Thx

[joachimwolff]

Why do you think it works only for h5?! It works for the cool-format of course too. Whole HiCExplorer supports for ALL its tools the cool format natively. Normalization is to adjust for different read coverages. If you don’t have multiple matrices, this doesn’t matter. Davide ***@***.***> schrieb am Do. 23. Jan. 2025 um 15:03:

…

Cooler does not scale the values back. The KR correction transforms all values to the 0-1 range. In HiCExplorer we expect the data to be in the original data range. If you use our hicCorrectMatrix, we scale the values back. Absolutely no need to use cooler for this. Davide *@*.***> schrieb am Do. 23. Jan. 2025 um 14:10: … <#m_-375354367464341630_> Thank you. So, since hicCorrectMatrix only works on h5 files, I guess I need first to convert raw hicpro matrices to .h5 then run hicCorrectMatrix applying the desired correction *and then convert them to .cool No need to use firs hicNormalize as suggested right? Thx — Reply to this email directly, view it on GitHub <#921 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADGQCAHVHLNEA62QHHD7RIT2MDZDZAVCNFSM6AAAAABVXKDS76VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDMMBZHA4TANZYGM> . You are receiving this because you commented.Message ID: ***@***.***>

[davidecarnevali]

Why do you think it works only for h5?! It works for the cool-format of course too. Whole HiCExplorer supports for ALL its tools the cool format natively.

Beacuse that's what the documentation says: "--matrix, -m: Name of the Hi-C matrix to correct in .h5 format."

Normalization is to adjust for different read coverages. If you don’t have
multiple matrices, this doesn’t matter.

Got it.
Thank you

Davide @.***> schrieb am Do. 23. Jan. 2025 um 15:03:
…

[joachimwolff]

I see, so we have to update our document on this. Thanks for pointing it out. Davide ***@***.***> schrieb am Do. 23. Jan. 2025 um 22:21:

…

Why do you think it works only for h5?! It works for the cool-format of course too. Whole HiCExplorer supports for ALL its tools the cool format natively. Beacuse that's what the documentation says: "--matrix, -m: Name of the Hi-C matrix to correct in .h5 format." Normalization is to adjust for different read coverages. If you don’t have multiple matrices, this doesn’t matter. Got it. Thank you Davide *@*.***> schrieb am Do. 23. Jan. 2025 um 15:03: … <#m_-5644278384061320343_> — Reply to this email directly, view it on GitHub <#921 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADGQCAHLVC43QUPVCZYAMKT2MFMMDAVCNFSM6AAAAABVXKDS76VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDMMJRGA2DCMRVGI> . You are receiving this because you commented.Message ID: ***@***.***>