Hi all,
I've successfully run RepeatMasker with hard mask and soft mask parameters and have been asked to pull out sequences which have masked repeat elements.
I'm new to the command line and can only use grep, awk, etc. in a very basic way.
Would anyone be able to help me with this?
Thanks!
Since I'm working on a cluster and don't have bedtools installed or privileges to install it (either on the cluster or on my local machine) I came up with this work around:
1. Change the space-separated .out file into a tab-delimited file:
cat FILE.out | tr -s ' ' | sed 's/^ *//g' | tr ' ' '\t' > FILE.out.tab
2. Extract the fifth column with sequence names, get rid of the duplicates, then cut first three lines out (these are there from making the file tab-delimited):
cut -f5 FILE.out.tab | sort -u | tail -q -n +4 > repeat.sequence.names.list
3. Make your .masked file a one-line file for easier manipulation (you do have to type > sign in the second line):
sed '/>/s/$/</g' < FILE.masked | tr -d '\n' | tr '<' '\n'| sed 's/>/\
>/g' | grep . > FILE.masked.1
4. Use the one-line .masked file to pull out sequences with repeats:
grep -A1 -f repeat.transcripts.list FILE.masked.1 | grep -v "^--$" > masked.sequences.repeat.fasta
There's an option of using grep with multiple CPU cores if you have parallel installed. See how to here.
Since you mention being familiar with awk:
awk '{OFS="\t"}{print $6, $7-1, $8}', though you should check include the strand).bedtools getfasta with a fasta file and the BED file from step one.I'm sorry, what do you mean by checking the strand? Thanks!
By the way, RepeatMasker's .out file isn't a tab-delimited file, unfortunately. Otherwise I reckon I could cut and sort -u the .out file that contains the names of sequences with repeat elements and grep those from my .masked file :/
That's the benefit of awk, it'll handle the fixed-width nature of the file (granted, you could fix this with sed too).
I used you kindly provided awk command on my .out file and ended up with this:
http://postimg.org/image/k0kt25hg7/
Which doesn't have the info I need :/ But I came up with another work around, so it's all good.