JAMM Parameter Recommendation

This page has what we think could be the best parameters to use with JAMM for some dataset types. It is possible though that for your particular case, other parameters could be better. Parameters provided here are just provided as examples and additional documentation.

If you have any questions or need help deciding on which parameters to use, please check the documentation or email us at this email and we'll be happy to help.

ChIP-Seq, Transcription Factor

The default JAMM parameters shoud be fine for transcription factor ChIP-Seq

ChIP-Seq, Punctate Histone Modifications

The default JAMM parameters shoud be fine for punctate histone modification ChIP-Seq. This includes, but is not limited to, H3K4me3, H3K4me2, H3K27ac, H3K9ac...etc.

ChIP-Seq, Broad Histone Modifications

Broad histone modification ChIP-seq is somewhat tricky mainly because they don't all have the same "broadness". Even the same antibody in different hands or different cell lines produces very different datasets in terms of resolution. In general, the following parameters can produce good results:

-r region -b 200

or

-r window -b 200

In fact, the -b parameter (which defines the bin size JAMM checks for enrichment) can go up to 10000 or even higer for very broad datasetes. For broad datasets, this is really a key parameter. We often test various bin sizes and select the best one. By default JAMM tries to estimate a plausible bin size but if your dataset is broad, bin sizes less than 200 do not really make much sense.

when you use large values for -b, it's better to set -w to 1:

-r window -b 10000 -w 1

DNase-Seq

The main point with DNase-Seq is to use the 5' ends of the reads. To do so use the -f parameter.

-f 1

If you have multiple replicates, JAMM expects a fragment length for each one of them, separated by commas. For example, for two replicates:

-f 1,1

ChIP-Seq, PolII

Just like broad histone modification ChIP-Seq, PolII is tricky. In some datasets, it looks like puctate transcription factors, in others it looks like broad histone modifications and sometimes it is even a mix of both.

So my advice is: look at your dataset in the genome browser. If it looks like transcription factors (there are well-defined peaks around transcription start sites) use default JAMM parameters. If it looks broad, use the broad histone modification settings above. If it's a mix of both I would recommend something like:

-r region

or

-r window

You can test different bin sizes if you think the one estimated by JAMM is not giving you the results you want.

ATAC-Seq

ATAC-Seq comes in two flavors in literature: paired-end and single-end. In the single-end case, it's just like DNase-seq (see above).

In the paired-end case, you will need to do some preprocessing before calling peaks. This is explained on this page. If after reading that page you still have any questions, feel free to email and ask.