Are eigenvector values from juicer_tools flipped?

180 views
Skip to first unread message

Azam Mohsin

unread,
May 9, 2020, 5:27:18 PM5/9/20
to 3D Genomics
When I run java -jar juicer_tools.jar eigenvector , I get positive and negative values. I assumed the positive corresponded to compartment A and the negatives correspond to compartment B. However, this contradicts my previous analysis. It seems that I have to flip the sign on all values, and then positive will be compartment A and negative will be compartment B.
Is the output from java -jar juicer_tools.jar eigenvector flipped?

Muhammad Saad Shamim

unread,
May 9, 2020, 8:08:34 PM5/9/20
to Azam Mohsin, 3D Genomics
Hey Azam,

The sign itself is arbitrary and can be flipped, utilizing other tracks to determine which way the flip should be.

On Sat, May 9, 2020, 4:27 PM Azam Mohsin <aamo...@uchicago.edu> wrote:
When I run java -jar juicer_tools.jar eigenvector , I get positive and negative values. I assumed the positive corresponded to compartment A and the negatives correspond to compartment B. However, this contradicts my previous analysis. It seems that I have to flip the sign on all values, and then positive will be compartment A and negative will be compartment B.
Is the output from java -jar juicer_tools.jar eigenvector flipped?

--
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/77386f69-00b1-421f-80b1-2d011f792d61%40googlegroups.com.

Azam Mohsin

unread,
May 9, 2020, 10:32:33 PM5/9/20
to 3D Genomics
Sorry, I am a little confused. Do you mean another hic track file? How would I determine the compartments for that file then? Is there no way to find A and B compartments for a hic file without using another hic file? Shouldn't these compartments be independent across files?

Thanks,

Azam

On Saturday, May 9, 2020 at 7:08:34 PM UTC-5, Muhammad Saad Shamim wrote:
Hey Azam,

The sign itself is arbitrary and can be flipped, utilizing other tracks to determine which way the flip should be.

On Sat, May 9, 2020, 4:27 PM Azam Mohsin <aamo...@uchicago.edu> wrote:
When I run java -jar juicer_tools.jar eigenvector , I get positive and negative values. I assumed the positive corresponded to compartment A and the negatives correspond to compartment B. However, this contradicts my previous analysis. It seems that I have to flip the sign on all values, and then positive will be compartment A and negative will be compartment B.
Is the output from java -jar juicer_tools.jar eigenvector flipped?

--
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-ge...@googlegroups.com.

Muhammad Saad Shamim

unread,
May 10, 2020, 12:31:01 AM5/10/20
to Azam Mohsin, 3D Genomics
From Hi-C data you can calculate an eigenvector - but this is a property of any matrix and the sign of such a vector is arbitrary. So you can always multiply it by -1 and flip the sign. To know what the correct sign is, you need additional information beyond Hi-C data (knowledge of which genes are active, methylation, etc.) and you flip the sign to correlate with this signal.


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/a28e4d76-4c4e-439a-9ccb-b6ef7876e8ad%40googlegroups.com.

Shuangyi Xu

unread,
Aug 31, 2021, 3:40:31 PM8/31/21
to 3D Genomics
Hi I'm just wondering if the sign can be flipped for one chromosome and not flipped for another? Is the arbitrary sign universal for one Hi-C file or it could just based on one chromosome? Thank you!

Muhammad Saad Shamim

unread,
Aug 31, 2021, 4:35:49 PM8/31/21
to Shuangyi Xu, 3D Genomics
Each chromosome is independent. 

Unless you want to use the new script which takes care of this issue:

Reply all
Reply to author
Forward
0 new messages