RSAT : peak-motifs

RSAT - peak-motifs

Discover exceptional motifs (over-represented, positionally biased) in a collection of ChIP-seq peaks.

References

Thomas-Chollier, M., Herrmann, C., Defrance, M., Sand, O., Thieffry, D. and van Helden, J. (2011). RSAT peak-motifs: motif analysis in full-size ChIP-seq datasets Nucleic Acids Research doi:10.1093/nar/gkr1104, 9. [Open access]
Thomas-Chollier M, Darbo E, Herrmann C, Defrance M, Thieffry D, van Helden J. (2012). A complete workflow for the analysis of full-size ChIP-seq (and similar) data sets using peak-motifs. Nat Protoc 7(8): 1551-1568. [PMID 22836136]

Information on the methods used in peak-motifs

The idea behind peak-motifs is that we detect exceptional words based on distinct and complementary criteria:

global over-representation (oligo-analysis and dyad-analysis): a word/dyad is more frequent than expected from the background model. The over-repressentation is tested with a right-tailed binomial significance test.
positional bias (position-analysis): a word has a heterogeneous of occurrences in the test sequences, i.e. there are regions with higher frequency and other regions with lower frequencies than the average of the same word observed over the whole width of the sequences. Positional bias is tested with a chi-squared tests.
local over-representation (local-word-analysis): the same test as for oligo-analysis (significance of the right tail of the binomial distribution) applied successfully to positional windows defined over the test set.

For position-analysis and local-word-analysis, the sequences are supposed to be aligned over some reference. For peaks, the reference is the summit (or center) of each sequence.

Peak Sequences

Title (mandatory)

Peak sequences (mandatory) Paste your sequence (fasta format)

Or select a file to upload (.gz compressed files supported)

URL of a sequence file available on a Web server (e.g. Galaxy).

Mask

Optional: control dataset for differential analysis (test vs control)

Control sequences Paste your sequence (fasta format)

Or select a file to upload (.gz compressed files supported)

URL of a sequence file available on a Web server (e.g. Galaxy).

Mask

(I only have coordinates in a BED file, how to get sequences ?)

Reduce peak sequences

Restrict the test dataset Number of top sequences to retain
Cut peak sequences: +/- bp on each side of peak centers

Motif discovery parameters

Discover motifs

Oligonucleotides (k-mers)

Discover over-represented words

[oligo-analysis]

Discover words with a positional bias

[position-analysis]

Discover words with local over-representation

[local-word-analysis]

Note:position-analysis and local-word-analysis will not run if a control set is provided

Oligomer lengths for the three programs above 6 7 8 merge lengths for assembly
Note: motifs can be larger than word sizes (words are used as seed for building matrices)

Markov order (m) of the background model for oligo-analysis (k-mers) (only for single-dataset analysis, will be ignored if control set is provided)

Spaced word pairs (dyads)

Discover over-represented spaced word pairs

[dyad-analysis]

Number of motifs per algorithm
Search on

Compare discovered motifs with databases (e.g. against Jaspar) or custom reference motifs

Compare motifs

Compare discovered motifs with known motifs from databases

Input

1 - Select a database in list:

2 - Select a collection in list:

View matrix descriptions & download full collections

Add your own motif database:

Add known reference motifs for this experiment:

Database and reference motifs (matrices) should be in Transfac format
(other formats can be converted with convert-matrix).

Locate motifs and export predicted sites as custom UCSC tracks

Locate motifs Search putative binding sites in the peak sequences [matrix-scan]
Markov order (m) of the background model for sequence scanning (site prediction + motif enrichment)

Visualize peaks and sites in genome browser No
Peak coordinates specified in fasta headers in bedtools getfasta format (also for retrieve-seq-bed output).
       Fasta headers should be in the form: >3:81458-81806(.)
Peak coordinates specified in fasta headers of the test sequence file (Galaxy format)
       Fasta headers should be in the form: >mm9_chr1_3473041_3473370_+
Peak coordinates provided as a custom BED file.
The 4th column of the BED file (feature name) must correspond to the fasta headers of sequences
           Assembly version (UCSC)

Reporting options

Plotting options Origin Offset

Output displayemail
Note: email output is preferred for very large datasets or many comparisons with motifs collections

[MANUAL]

[TUTORIAL]