search

robinweide / GENOVA

GENome Organisation Visual Analytics
GNU General Public License v3.0
68 stars 15 forks source link

PEScan CTCF plot from the paper #321

Open odovgusha opened 1 year ago

odovgusha commented 1 year ago

Hi, thank you for the package.

I struggle to get CTCF and RAD21 aggregation plots for my human ESC data. So, far I only manage to get PEscan plots only for PRC2 Chip-seq peaks using default settings dist_thres = c(5000000L, Inf). These settings do not work for the CTCF and RAD21. It seems that dist_thres plays an important role in aggregation plots.

Could you please share the settings which you had in your paper to aggregate the interactions between CTCF Chip-seq peaks (Figure 2. F) and also suggest which settings should I use for RAD21?

For HiCExplorer tutorial for contact aggregation, it was suggested to keep the minimum distance over the average size of TAD. This works for PRC2 data but I guess for CTCF I might use a different threshold. I did not find a proper guide on the web for this.

Many thanks!

teunbrand commented 1 year ago

@robinweide I have no clue what exact parameters you used here. Do you recall this?

odovgusha commented 1 year ago

The plots which I get for CTCF (default settings) image (dist_thres = c(400000, 3000000)) image For PRC2 (default settings) image

odovgusha commented 1 year ago

@robinweide I have no clue what exact parameters you used here. Do you recall this?

I do not need the same exact settings. I struggle to understand how should one change the settings in GENOVA depending on which ChIP-seq data is used (CTCF/RAD21 or RING1B). Especially, shift and distance threshold. It seems that for RING1B one should use bigger thresholds c(2000000,20000000). While for CTCF we are looking for smaller-scale interactions. c(200000,1000000).

I also tried to apply PEscan for published data and have problems with getting a proper aggregation plot for CTCF as well. Many thanks!

Best regards, Oleksandr