Query regarding FIMO input

[Sangram keshari Sahu]

I am trying to run FIMO for all the motifs collected from different sources (Only for Arabidopsis) in the form of PWM (in meme format) against my interest promoter region. All the PWM have different "background frequency".

I want to know does "background frequency" matters for FIMO input and if yes how to calculate them for single PWM input in meme format to FIMO . 

[CharlesEGrant]

The background model is a key input for FIMO. FIMO scores motif matches by evaluating the log ratio of the likelihood of a sequence assuming the motif position specific frequencies and the likelihood assuming the background frequencies. FIMO’s default behavior is to use background frequencies derived from the Non-redundant (NR) database. This is the only option currently available for the FIMO web application when using user uploaded sequences. For scans of the provided on-line sequence databases in the web application, FIMO will use a pre-computed background, specific to those sequences.

If you are using the command line version of FIMO you can use the ‘--bgfile’ option to specify a custom background model. The FIMO command line options are described here: 

http://meme-suite.org/doc/fimo.html.

The ‘--bgfile’ option requires a additional value, which is usually the name of the file containing the background frequencies. The format of the background file is described here: 

http://meme-suite.org/doc/bfile-format.html. 

FIMO only supports a 0th order background model. 

There are a couple special values for the ‘--bgfile’ option. '--motif--‘ directs FIMO to use the background frequencies specified for the motif in the input MEME file. ‘--uniform--‘ which directs FIMO to use a uniform background model, with equal probabilities for each letter in the core alphabet.

You can create a background file from your sequence data by using the utility ‘fasta-get-markov’, which is described here:

http://meme-suite.org/doc/fasta-get-markov.html

Ideally the background model should be derived from sequences that are biologically similar to the sequences you are scanning, but which don't contain instances of the motifs you are scanning for. However, simply running 'fasta-get-markov’ on the sequences you are going to scan with FIMO is a decent starting point.