Batch effect

[Shirlekar, Kalyanee]

Hello!

I am new to the tool and have got my hands on the HIC data generated in two batches. The documentation for juicer and juicebox has been a big help! Thank you!

I have a question about batch effect: Here is my experiment information. Each sequencing batch has 2 conditions, WT and KO (2 biological replicates each), i.e., 4 samples in batch 1 and 4 samples in batch 2. I generated individual .hic files using juicer for the 8 samples. I was wondering if there is a way to check for batch effect and correct for it. It seems useful to know this before merging the biological replicates. 

Any ideas are welcome. 

Thank you. 

Kalyanee

[Moshe Olshansky]

Hi Kalyanee,

This is an interesting and important question to which I don't know the answer.

I would start by checking the consistency between the two replicates in one batch and if they are I would merge them and then check the consistency between the two batches. Except for visual inspection with juicebox, the only way I am aware of is calling loops and comparing the lists either with juicer_tools hiccupsdiff or in some other way (like how many of the top 1000 loops in one list overlap with loops in the second list and what their ranks are).

If everything is consistent, you can do what you want. If not, you will need to think what kind of analysis you want to do and how to deal with batch effect.

I am very much interested in other people's thoughts.

Best regards,

Moshe.

[Shirlekar, Kalyanee]

Hi Moshe,

Thanks for your suggestions. I looked at the 4 .hic files visually per condition (condition- KO), and unfortunately, the two replicates in batch 1 do not appear to be similar. They have different regions/domains. I talked to my collaborator and found out that the second batch was sequenced because of the discrepancy within the first batch. Visually, the replicates in the second batch of sequencing are appearing more consistent within each condition. We suspect that one of the replicates of KO in my first batch had some quality issues. I want to probe more into its QC before removing that replicate from my analysis completely. 

If you have any recommendations, please let me know. 

Thanks again!

-Kalyanee