What is the Eigenvector input .hic file?

[Jenny]

Hi all,

I am trying to get A/B compartments from my Hi-C library.

I have generated .hic files using juicer.sh and I also have used juicer_tools.jar pre to generate a 40kb resolution .hic file.

Then I tried the following:

for i in {1..22}

do

    java -jar $Juicer_tools eigenvector KR $hic_file $i BP 40000 $HOME/Eigenvector/${sample_name}_chr${i}_eigen.txt

done

java -jar $Juicer_tools eigenvector KR $hic_file X BP 40000 $HOME/Eigenvector/${sample_name}_chrX_eigen.txt

java -jar $Juicer_tools eigenvector KR $hic_file Y BP 40000 $HOME/Eigenvector/${sample_name}_chrY_eigen.txt

However, all the output files are 0B.

The Eigenvector.oxxx file only says: WARNING: Pearson's and eigenvector calculation at high resolution can take a long time

The Eigenvector.exxx file says:

Unknown resolution: BP_40000

This data set has the following bin sizes (in bp): 

2500000 1000000 500000 250000 100000 50000 25000 10000 5000 

and the following bin sizes (in frag): 

500 200 100 50 20 5 2 1

When I changed to use .hic file from juicer_tools.java pre, I got only the warnings but no error message. However, all the output files are still 0 byte.

I'm wondering what shall I do to make this work?

Thanks,

Jenny

[Moshe Olshansky]

Hi Jenny,

I have an R script to compute the eigenvector at high resolution (even 5 kb resolution for mouse chr1 takes a few minutes or less).

I think that you can get it from juicer github. If not I can e-mail you my scripts.

Moshe.

[Neva Durand]

Jenny,

There are several issues. 

1) You did not create a hic map with the resolution 40000 so you can’t do an eigenvector at that resolution. 

2) You need to include the -p flag since you are trying to calculate an eigenvector at very high resolution. It can take a while. 

Moshe has scripts for this however I’m not sure they operate on the same matrix (the O/E is whitened before calculating Pearson, creating a dense matrix, which is why it is a very long calculation). 

Best

Neva

--
You received this message because you are subscribed to the Google Groups "3D Genomics" group.
To unsubscribe from this group and stop receiving emails from it, send an email to 3d-genomics...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/3d-genomics/a6397099-e18f-4d51-b821-7ab64fe05eb6%40googlegroups.com.

--

Neva Cherniavsky Durand, Ph.D.

Staff Scientist, Aiden Lab

www.aidenlab.org

[Jiani Yin]

So I can start with the 40kb .hic file from juicer_tools pre, and then add -p flag to the juicer_tools eigenvector. Let me try this first.

Thank you guys.

Jenny

To view this discussion on the web visit https://groups.google.com/d/msgid/3d-genomics/CAF1CciWCbkU9m-7gWE5Lm-Dk3q0SLS0wLLDWhHE0q-KvA7kGEg%40mail.gmail.com.

[Moshe Olshansky]

Yes. And at 40 kb resolution it should not take more than an hour or two for mouse or human chr1 and much less time for smaller chromosomes.

On Friday, August 16, 2019 at 12:19:29 PM UTC+10, Jenny wrote:

[Jiani Yin]

You are right, it took 40-50min for the X chromosome.

I’m thinking is there more value to do this on 40kb resolution map? Or 100kb is sufficient for compartment level analysis?

Jenny

--
You received this message because you are subscribed to the Google Groups "3D Genomics" group.
To unsubscribe from this group and stop receiving emails from it, send an email to 3d-genomics...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/3d-genomics/a43a0885-509e-475b-b353-f9cb754d5240%40googlegroups.com.

[Moshe Olshansky]

I think that you can do this at 100 kb resolution but Neva should be able to answer this better than me.

We were using this for TADs and wanted to try 5 kb resolution which would take more than a month for mouse chr1 and this is what made me look for an alternative.

On Friday, August 16, 2019 at 12:19:29 PM UTC+10, Jenny wrote:

[Neva Durand]

Yes, for compartments 100Kb is fine. 

--
You received this message because you are subscribed to the Google Groups "3D Genomics" group.
To unsubscribe from this group and stop receiving emails from it, send an email to 3d-genomics...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/3d-genomics/26cda245-42ce-4290-b931-f878eeced1c1%40googlegroups.com.

[Joji Marie Teves]

Hello Moshe, 

I'm interested to get the R script for high resolution eigenvector, is it possible to have it too? 

Thanks,

Joji

[Nasreen Bano]

Hello Moshe, 

I'm also interested to get this R script for high resolution eigenvector. Will you please give it to me?

Thanks,

Nasreen

[Moshe Olshansky]

Hi Nasreen,

Sorry for the delay - I somehow didn't notice your email.

The code is in 

https://github.com/moshe-olshansky/EigenVector

It contains both R and C/C++ code.

Please keep in mind that R code (even though it works) is older, slower and less flexible than the C/C++ code.

Let me know if you need further help.

Best regards,

Moshe.

--
You received this message because you are subscribed to a topic in the Google Groups "3D Genomics" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/3d-genomics/8dp9EChj4Nw/unsubscribe.
To unsubscribe from this group and all its topics, send an email to 3d-genomics...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/3d-genomics/f5899b8d-139f-42cc-95e3-49877eebc9e8n%40googlegroups.com.

[Nasreen Bano]

Thank you so much Moshe.

You received this message because you are subscribed to the Google Groups "3D Genomics" group.

To unsubscribe from this group and stop receiving emails from it, send an email to 3d-genomics...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/3d-genomics/CAFuhK5ycNurU76y4g8mOsgPHfTCyPwfQ7jy7UG%2BywTfNq6rPFA%40mail.gmail.com.