I am trying to a recently published software called diffTF and contacted the authors on how to obtain the input HOCOMOCO files for GRCh38. While I wait for the answer, I thought I would post the question here as well:

The tool requires bed files for each of the transcription factors (TF) in the genome, such as this bed file (TAB separated), which seems to give the locations of different TF binding sites in the genome:

chr1    3457132     3457141     GTTGCGTGA   1266   +
chr1    4595374     4595383     GTTGCGTGA   1266   -
chr1    4996631     4996640     TTTGCGTGC   1276   +
chr1    5005357     5005366     TTTGCGTGC   1276   +
chr1    5170456     5170465     TTTGCGTGC   1276   -
chr1    5567302     5567311     GTTGCGTGA   1266   -
chr1    8676279     8676288     TTTGCGTGC   1276   -
chr1    8859719     8859728     GTTGCGTGA   1266   +
chr1    9600920     9600929     GTTGCGTGA   1266   +
chr1    9650971     9650980     GTTGCGTGA   1266   +

Then there is a symbols file to map the TF to their gene(?) that looks like this:

Ahr     ENSMUSG00000019256      AHR
Aire    ENSMUSG00000000731      AIRE
Alx1    ENSMUSG00000036602      ALX1
Ar      ENSMUSG00000046532      ANDR
Tfap2a  ENSMUSG00000021359      AP2A
Tfap2b  ENSMUSG00000025927      AP2B
Tfap2c  ENSMUSG00000028640      AP2C
Tfap2d  ENSMUSG00000042596      AP2D

I googled for HOCOMOCO and found a bunch of pwn or words files, but none of them seems to be similar to the bed file above. Any ideas where to get these for GRCh38?

  • $\begingroup$ What have you tried besides contacting the authors? I am not familiar with this type of files, but it doesn't seem to be too strange, so perhaps they assumed it is fairly easy to download and transform from other files into these... $\endgroup$
    – llrs
    Feb 27, 2019 at 8:11

1 Answer 1


I guess you're right that you need predictions of transcription factor binding site (TFBS) through genome. There're two options to get them:

  1. You can generate predictions using any motif scanning tool. E.g. you can use our tool SPRY-SARUS which is able to generate bed-file in necessary format. In this case, you can choose more or less strict thresholds for each prediction (i.e. more false positives vs more false negatives). Standard thresholds for HOCOMOCO motifs can be found at its downloads page (we, as HOCOMOCO authors, recommend thresholds corresponding to P-value of 0.0005).
  2. I had no experience with diffTF but it seems that authors prepared TFBS predictions and put them on their site. Please, take a look at this page of manual. If you don't need extra-flexibility in the generation of TFBS predictions, it can be the simplest way to proceed.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.