Entering edit mode
Paul Shannon
▴
750
@paul-shannon-5161
Last seen 10.3 years ago
I offer below a short example of using rGADEM with MotifDb which you
may find useful.
The example motivates and then demonstrates "seeded" search, in which
a candidate TF is provided to rGADEM. (The rGADEM vignette
illustrates the simpler task of de novo motif search.
- Paul
On Feb 27, 2013, at 9:03 AM, Zhu, Lihua (Julie) wrote:
> Jose,
>
> Thanks for your kind word!
>
> You might want to check out rGADEM package for de nova motif
discovery.
>
> Best regards,
>
> Julie
>
library(rGADEM)
library(MotifDb)
library(BSgenome.Hsapiens.UCSC.hg19)
path <- system.file("extdata/Test_100.fasta", package="rGADEM")
sequences <- readDNAStringSet(path, "fasta")
length(sequences) # 49
# you can use the UCSC genome browser to BLAT the first 4
sequences, one at a time,
# against human hg19, and see what TFs bind there
# http://genome.ucsc.edu/cgi-bin/hgBlat?command=start
as.character(substr(sequences[1:4], 1, 80))
# Once you blat, click through to view the blat match in the genome
browser,
# and enable the "Transcription Factor ChIP-seq from ENCODE" track.
# The FOXA1 Tf has a high score in this region of chr1
# for most of these sequences
# obtain the reported motif for this TF from the Bioc MotifDb
package
query(MotifDb, 'foxa1')
# MotifDb object of length 1
# | Created from downloaded public sources: 2012-Nov-01
# | 1 position frequency matrices from 1 source:
# | JASPAR_CORE: 1
# | 1 organism/s
# | Hsapiens: 1
# Hsapiens-JASPAR_CORE-FOXA1-MA0148.1
# now use rGADEM to do a seeded search:
pwm.foxa1 <- as.list(MotifDb["Hsapiens-JASPAR_CORE-FOXA1-MA0148.1"])
gadem.foxa1
<-GADEM(sequences,verbose=1,genome=Hsapiens,Spwm=pwm.foxa1, seed=TRUE)
consensus(gadem.foxa1)
# [1] "nTGTTTACwyw" "GmwGrrrsswGGvAGn"
# how do the 49 sequences match to these two motifs?
mean(sapply(gadem.foxa1 at motifList[[1]]@alignList, function(x) x at
pval))
mean(sapply(gadem.foxa1 at motifList[[2]]@alignList, function(x) x at
pval))
On Feb 27, 2013, at 9:03 AM, Zhu, Lihua (Julie) wrote:
> Jose,
>
> Thanks for your kind word!
>
> You might want to check out rGADEM package for de nova motif
discovery.
>
> Best regards,
>
> Julie
>