shifting reads in cisgenome for chip-seq 2-sample analysis

[bbl]

Hello there,

I am following the cisgenome online tutorial for 2-sample chip-seq analysis. I compare the median of peak length calculated from hts_peakdetectorv2_2sample -br 0 -ssf 0 and -br 1 -brl 30 -ssh 1.

The difference of median is very large: 199 and 88 bp. Is it normal? Can boundary refinement and single strand filtering affect the peak length so dramatically?

Another question following up is should we just use default value with no br and ssf or should we use the parms in the tutorial?

Thanks.

[bbl]

any suggestions from who have experience with CisGenome?

[zee]

I have used CisGenome before and I basically went with the tutorial recommendations for those parameters I did not understand too well. I do know it is crucial to ensure CisGenome knows the correct read length from your runs.
Otherwise it's good to mail the authors this type of question.

Have a go with the tutorial parameters and see if you can get the whole pipeline running, then start looking at how to tweak parameters and their effects.

[bbl]

It seems to me that how many bp to shift the reads are crucial for identifying the TFBS. Since different parameters give such a dramatic different L, I am not very confident for downstream TFBS identification without understanding this well.