Run Taiji on Cut&Run histone mark data

[helenhuangmath]

Hi Taiji author,

Thanks for creating this useful tool for key TF identification!

I have some general questions of running Taiji on RNA + histone marks.

1. I'm running RNA + H3K27ac bam files using Taiji, but got no significant TF identified, and no *html heatmap output file. What possible reasons do you think this could be? Is it possible that Cut&Run signals are more wider than ATAC that not fit well for Taiji? Is there any way to also add peak files in the config file or in the run to limit called peaks? If so, could you show examples or tutorial of doing so?

2. Would it be possible to run RNA + broad repressive histone marks such as H3K27me3 to referring repressive TF using Taiji? Do you have any thoughts or suggestions on it?

Thanks for your help and suggestions!

[kaizhang]

@helenhuangmath Thanks for using Taiji.

1. TFs that are not present in *.html are not necessarily insignificant. You should look at and analyze the 'GeneRanks.tsv' file to identify significant TFs. It's possible to use custom peak files, see here: https://taiji-pipeline.github.io/documentation/input.html#atac-seq

2. I haven't tried that, but I'm afraid that won't work smoothly. I'm now moving towards single-cell data to address this type of questions.

[helenhuangmath]

Hi Kai,

Thank you for your reply!

Continuing on question 1, Is it possible to give Taiji both bam file and peak file to allow it using both enrichment level information from bam and pre-defined peak information? Or it can take only one type, either bam or peak?

Thanks again!

[kaizhang]

Yes, it can use both.