Hi,
What are some popular methods that go beyond using position weight matrices (PWMs) to determine putative transcription factor binding sites (TFBS)? The rate of false positives is very high so I am looking for better ways to assess the probability that a sequence is indeed a binding site.
I am looking at the promoters of a set of genes (2.5k region upstream of the transcription start site) and separately at creating a list of potential targets for certain transcription factors. I am using MotifDb to obtain PWMs for TFs in Mouse and TFBSTools to search for possible TFs that would bind in the promoter regions.
I am not interested in using pipelines such as MEME and others which leave little flexibility, I would much rather read about a method and implement it. However, I am unsure as to which would be a reliable one which would yield better results than just using PWMs.
Thank you for you help!
Integrating high-resolution ChIP-seq data might be useful for TFBS discovery. There's a paper here that discusses one approach.
Thank you very much!
Unfortunately there is limited Chip-seq data available for this particular case (types of stem cells) and the purpose of the analysis I aim to undergo is to reveal other potential transcription factors which might be important and therefore create a basis for future Chip-seq experiments.
While the use of PWMs is still very much the standard, there has been some recent work in using alternate approaches.
You may find the recent work of Mathelier and Wasserman (PLOS CB, 2014), who employ HMMs to extend the predictive capabilities of PWMs to be of interest in this context.
Log in to answer this question.
