I have been processing some ChIP-seq data with the R package spp. I looked through the literature (ENCODE uses it) and it seemed that spp is indicated as a good program to use. I have found and adapted the two tutorials I found for using spp (first and second), and read the original paper. I have also e-mailed Prof. Karchencko and posted on the bioconductor listserv - all with no response. My question is about the MSER and the predicted sequencing depth - additional output of spp. So what I think the MSER is, is the score value above which a peak is authoritatively determined? When I look at peaks above this score value they are very well defined. Also, is the predicted sequencing depth is a prediction of the number of additional "tags" necessary so that the MSER and the FDR value coincide? It is hard to find decent information about this program. Any advice or additional information is greatly appreciated!


  • $\begingroup$ does the inline spp documentation(> ??spp) explain it any better? $\endgroup$
    – MattDMo
    Commented Mar 19, 2013 at 17:30
  • $\begingroup$ I found the answer in this paper, ncbi.nlm.nih.gov/pmc/articles/PMC3191340 $\endgroup$ Commented Mar 19, 2013 at 20:36
  • $\begingroup$ especially this figure,ncbi.nlm.nih.gov/pmc/articles/PMC3191340/figure/F3 $\endgroup$ Commented Mar 19, 2013 at 20:36
  • $\begingroup$ There are no inline spp docs. $\endgroup$ Commented Mar 19, 2013 at 20:39
  • $\begingroup$ if you type in ??spp at the R prompt you'll get the docs. They're not expansive, but they explain the different parts of the library. $\endgroup$
    – MattDMo
    Commented Mar 19, 2013 at 21:55

1 Answer 1


I found the answer to this question and some more interesting details that people who stumble upon this might also need. First look at this paper, http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3191340/ and especially this figure, http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3191340/figure/F3/. The MSER is a ratio of the fraction of discovered peaks to the fraction of the sampled reads to recapitulate them, so if you are able to return all of the peaks with a small sample of reads then you have excellent coverage. Now even though you have great and well defined peaks you need another sample to confirm that they are indeed significant, and you can do this with the IDR, see this website - https://sites.google.com/site/anshulkundaje/projects/idr. The IDR is laid out pretty well in the supplemental material for this ENCODE paper, http://www.nature.com/nature/journal/v489/n7414/full/nature11245.html. Beautiful!!


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .