FourCSeq - getZScores & getDifferences function
Entering edit mode
mqbsszsy • 0
Last seen 6.0 years ago
United Kingdom

Hi Felix/Wolfgan,

Thank you very much for the previous email correspondence (see below) on FourCSeq package. 

>Question: How do I bypass the counting procedure and precede the downstream statistical analysis in detecting interactions and differences?

>Your answer: You have to create the FourC Seq object as described in the vingette with colData and exptData. Then you skip all steps as digesting the reference and couning read. But you have to generate your GRanges object of the fragments yourself.


I have further questions regarding to the data analysis.

(1). In “getZScores” function: Prior to obtain VST count, the raw count is normalised by the sum of filtered read count (such as, minCount=40) but not total sequenced reads. Is it correct?

(2). How do we evaluate whether “distFitMonotoneSymmetric or distFitMonotone” is the choice of fitting? In addition, could you please advise me on how to perform the plotFit on “distFitMonotone” data (no issue with distFitMonotoneSymmetric using “plotFits(fcf[,1]" command)?

(3). In “getDifferences” function, it is not clear to me whether the calculation of normalisation factors is simply based on the raw count (back-transformed trend fitted value) or filtered normalised count (see 1). If the answer is raw count, shouldn’t we obtain differential interactions based on count normalised with both distance and sequencing depth?

Thank you very much for your help

fourcseq • 634 views
Entering edit mode
felix.klein ▴ 150
Last seen 3.1 years ago
Hello, here are the answers to your question: (1) I dont know what you exactly mean by sum of filtered read count. For the calculation of the VST library size factors are calculated, but the raw signal is not normalized. For the details I would recommend you have a look at the formulas in our paper which describe it in the most concise way. It is freely available here: (2) a) Could you please send the code that you used and error that you obtained. Then I can traceback what went wrong. b) If you have very asymmetric profiles around your peak and want to take this into account you would use distFitMonotone. If this signal is symmetric or you want to assume a symmetric signal to detect asymmetry then you use distFitMonotoneSymmetri. (3) For this you should also have a look at the formula in the paper. Best regards, Felix

Login before adding your answer.

Traffic: 573 users visited in the last hour
Help About
Access RSS

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6