Hello,
I want to compile a BED file with genomic coordinates of all occurrences of a particular sequence motif in the complete genome. One obvious online tool to do this task, FIMO, can not be applied because it has a limit of 1 million nucleotides. I need to scan the whole genome (~1e9 bp). Any software suggestions?
Thanks!
3 answers
If I understand correctly, you can download and install MEME/FIMO locally (http://meme.nbcr.net/meme/meme-download.html, http://meme.nbcr.net/meme/doc/fimo.html). That might solve the 1Mbp limit. If it doesn't you could chop up your sequence like this:
bedtools makewindow -g GEN.fa -w 1000000 > GEN.bed
bedtools getfasta -fi GEN.fa -bed GEN.bed -fo GEN-win.fa
csplit -z GEN-win.fa '/>/' '{*}'
for FILE in GEN-win.fa.*; do FIMO; done
You would have to adjust coordinates by 1Mbp * split.count for each result file
Use TRAP. What you need is the genome in fasta format and the motif matrix. It is possible to use Transfac and Jaspar or use your own motif as a position-specific scoring matrix.
You may want to take a look at the Nature Protocols paper describing the tool.
Thank you. Could you please tell me an approximate duration of such a job? As you have suggested, I am running a command line with the same parameters as you just indicated. My motif length is 27 nucleotides, and the genome is human. It took already more than 3 hours, still running, and no indications of the state of the job...
24 hours is too much! The problem may lie on the --ttype balanced, to solve this replace it by -t 5.0 (I just asked my boss Thomas Manke who participated in the development of the tool). The -t (threshold) is the score used to define hits. Higher numbers result in fewer hits that are of higher confidence. How did you compute the pssm matrix?
The MEME output only include the frequency counts. You need a matrix that has log ratios summing up the probabilities of finding a particular base against a background. Try converting your meme matrix using this tool: http://rsat.ulb.ac.be/convert-matrix_form.cgi
We just tried and a local run took only 10 minutes using the human genome.
So finally I ended up mapping the fixed motif by Bowtie using the "-a" option. It took just several seconds.
FIMO did not install locally after several attempts and upgrades of the Perl and Python. Some compilation errors...
TRAP did install locally, but it did not finish the job, and the format of the input matrix required for TRAP remains unclear.
Log in to answer this question.
-
Find motifs relative location to features
written by Ruixuan •
-
How toconvert a set of motif consensus sequences in IUPAC to meme format?
written by ilovesuperheroes1993 •
-
Codis marker exact genomic coordinates
written by nmmsv •
-
Tutorial: Finding individual motif occurrences with FIMO from the MEME suite
written by ATpoint
-
MEME result v FIMO result for target annotation
written by BioinfGuru
-
Scan a genome for the ocurrence of a sequence motif in the form ...CC[CT]C[AC]GCCG[CT]CC...
written by qwerty •
-
Make motif position matrix based on multiple sequence alignment.
written by philrees •
-
Find all occurrences of a fixed sequence in a large genome
written by qwerty •
-
I Am Trying To Run Fimo From Meme Suite
written by k.nirmalraman
-
Calculating Background Model For Genome Scanning With Pwm
written by Amm •
What does your motif look like? Is it a sequence you guys are working on (i.e. not in databases already) or a previously known motif?
You could try IGV http://www.broadinstitute.org/igv/motif_finder
It's not a known TF motif, it's a novel motif determined from ChIP-seq