ARSA search protocol

Blast

# Run blast:
blastall -p blastp -d /mnt/project/pracstrucfunc12/data/big/big_80 -m 8 -b 10000 -i ARSA.fas > ARSA.blast

# find unique matches
cut -f 2 ARSA.blast | uniq > ARSA.blast.matchIDs.uniq
uniq -w 44 ARSA.blast > ARSA.blast.uniq

# use R to plot: 
# within R: histogram of log10 Evalue for blast run
blastTable = read.table("ARSA.blast.uniq")
pdf("ARSA_blastHistogram_eVal.pdf")
logeval.hist <- hist(log10(blastTable$V11), breaks=100, plot=FALSE) 
plot(logeval.hist, col="blue", main="BLAST logarithm of E-values", xlab="lg(E-Value)")
dev.off()

# within R: histogram of %identity for blast run
blastTable = read.table("ARSA.blast.uniq")
pdf("ARSA_blastHistogram_id.pdf")
id.hist <- hist(blastTable$V3, breaks=5, plot=FALSE)
plot(id.hist, col="blue", main="BLAST sequence identity distribution", xlab="percent identity")
dev.off()

 # create lists of Uniprot IDs for analysis for GO annotations
 grep -n "1e-10" ARSA.blast.uniq
 tail -n +1721 ARSA.blast.uniq > ARSA.blast.uniq.worseE10
 cut -f 2 ARSA.blast.uniq.worseE10  > ARSA.blast.uniq.worseE10.matchIDs
 cut -d "|" -f 2 ARSA.blast.uniq.worseE10.matchIDs > ARSA.blast.uniq.worseE10.matchIDs.clean

PSI-Blast

# run PSI-Blast with different combinations of parameters
> blastpgp -d /mnt/project/pracstrucfunc12/data/big/big_80 -m 8 -b 10000 -j 2 -i ARSA.fas -o ARSA.psiBlast.j2.hDefault
280.470u 30.420s 5:55.56 87.4%  0+0k 18005712+1880io 792pf+0w
> blastpgp -d /mnt/project/pracstrucfunc12/data/big/big_80 -m 8 -b 10000 -j 10 -i ARSA.fas -o ARSA.psiBlast.j10.hDefault
[blastpgp] ERROR: ncbiapi [000.000]  ObjMgrNextAvailEntityID failed with idx 2048
2111.750u 44.740s 36:55.25 97.3%        0+0k 12951072+11856io 1271pf+0w 
> blastpgp -d /mnt/project/pracstrucfunc12/data/big/big_80 -m 8 -b 10000 -j 2 -h 1e-10 -i ARSA.fas -o ARSA.psiBlast.j2.h1e-10
280.140u 6.580s 4:55.89 96.9%   0+0k 2536896+2000io 258pf+0w
> blastpgp -d /mnt/project/pracstrucfunc12/data/big/big_80 -m 8 -b 10000 -j 10 -h 1e-10 -i ARSA.fas -o ARSA.psiBlast.j10.h1e-10

# To evaluate the final results, we have to dissect the output into separate files: Look for the first hit, find at what line numbers it occurs and then cut the files accordingly.
grep -n G3IH84 ARSA.psiBlast.j2.hDefault
tail -n +4679 ARSA.psiBlast.j2.hDefault > ARSA.psiBlast.j2.hDefault.lastIter

# Extract unique matches as above for Blast
uniq -w 44 ARSA.psiBlast.j2.hDefault.lastIter > ARSA.psiBlast.j2.hDefault.lastIter.uniq

# generate histogram of e-value distributions in R:
#colors
cols4 <- hcl(h = seq(30, by=360 / 4, length = 4), l = 65, alpha = 0.5)
# read data
j2_hDefault = read.table("ARSA.psiBlast.j2.hDefault.lastIter.uniq")
j10_hDefault = read.table("ARSA.psiBlast.j10.hDefault.lastIter.uniq")
j2_h10 = read.table("ARSA.psiBlast.j2.h1e-10.lastIter.uniq")
j10_h10 = read.table("ARSA.psiBlast.j10.h1e-10.lastIter.uniq")
# hist1
hist(j2_hDefault$V2,border=cols4[1],col=cols4[1],breaks=151,main='PSI-BLAST logarithm of E-values',panel.first = grid(),xlab='lg(E-value)',axes=FALSE,xlim=range(-150,1),ylim=range(0,1000))
# hist2
hist(j10_hDefault$V2,border=cols4[2],col=cols4[2],breaks=151,add=TRUE)
#axes
axis(1, las=1, labels=seq(-150,0,50), at=seq(-150,0,50), tcl=0.5)
axis(2, las=1, labels=seq(0,1000,100), at=seq(0,1000,100), tcl=0.5)
#hist3
hist(j10_h10$V2,border=cols4[3],col=cols4[3],breaks=20,add=TRUE) 
#hist4
hist(j10_h10$V2,border=cols4[4],col=cols4[4],breaks=20,add=TRUE)
#legend
legend(0.8,100,c('j 2, h 0.002','j 10, h 0.002','j 2, h 1e-10','j 10, h 1e-10'),col=cols4,fill=cols4,border=cols4)