coordinates of all TFBS of human genome

[于振朋]

Hello, I confused about the coordinates of all TFBS of human genome.

I find the dataset in two resource: http://expdata.cmmt.ubc.ca/JASPAR/downloads/UCSC_tracks/2020/, and https://unibind.uio.no/static/data/20201216/bulk_Robust/Homo_sapiens/damo_hg38_all_TFBS.tar.gz.

I find the data of JASPAR is very big, more than 80G. I think this database may be redundant and not accurate prediction results.

In contrust, the data of Unibind is over 1G. It seems that it is experimentally verified.

So I want to ask if I should use the Unibind database instead of the JASPAR.

[于振朋]

By the way, the  Unibind  just have 268 TFs, which less that   JASPAR (1600+ TFs). 

If I want to get more information about transcription factor binding sites, I should deal with it. 

[Anthony Mathelier]

Hello,

The predictions in JASPAR and UniBind are made very differently.

For JASPAR, we took all PFMs in JASPAR and scanned the genomes with the same thresholds for all matrices. You can find the details in the supplementary material of the JASPAR 2018 manuscript (https://academic.oup.com/nar/article/46/D1/D260/4621338).

For UniBind, we took a more stringent/robust approach to predict high-quality direct TF-DNA interactions. The predictions require to have both computational (JASPAR PFMs) and experimental (ChIP-seq) evidence for direct TF-DNA binding. The necessity to have both PFMs and ChIP-seq available for a given TF reduces the amount of TFs for which we can make robust predictions. And the method finds the optimal PWM threshold to be used in a dataset-specific manner. You can find the details in the corresponding manuscript at https://www.biorxiv.org/content/10.1101/2020.11.17.384578.

I hope this addresses your questions.

--
You received this message because you are subscribed to the Google Groups "JASPAR Q&A Forum" group.
To unsubscribe from this group and stop receiving emails from it, send an email to jaspar+un...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/jaspar/85f1a146-45d2-48f3-bd4e-edae2a55345dn%40googlegroups.com.

-- 
Anthony

[zhenpeng yu]

Thanks for your help.