Thanks! It works; however, it is taking a lot of time to run. I will be searching hundreds of strings among thousands of files; and I suspect it will take way too long. This is the time I got when using your script in one real dataset searching for only one string:

Code:

real    0m28.877s
user    0m28.734s
sys     0m0.015s

Same dataset with my old script:

Code:

awk 'FNR==1{n=substr(FILENAME,1,index(FILENAME,".")-1)} { print n "\t" (/AATTCCGGATCG|CGATCCGGAATT/ ? "ATCG" : "NOT Present" ) }'

I got this:

Code:

real    0m0.074s
user    0m0.031s
sys     0m0.030s

It is pretty fast but it does not report the last 4 characters so it's no good.
I kinda get what I want using the following sed scripts:

Code:

sed -n 's/^[ATCG]*AATTCCGG\(.\)\(.\)\(.\)\(.\)[ATCG]*$/\1\2\3\4/p' file
sed -n 's/^[ATCG]*\(.\)\(.\)\(.\)\(.\)CCGGAATT[ATCG]*$/\4\3\2\1/p' file | tr ATGC TACG

The timing for the individual script, again using the same dataset mentioned above:

Code:

real    0m0.482s
user    0m0.405s
sys     0m0.138s

I would like to combine both sed script into one. Maybe using an IF statement in bash. Even though I would like to avoid bash if at all possible
I am also not sure how to modify so it can output the last three characters after the string

Almost there

Code:

 sed -n '0,/AATTCCGG/s/^[ATCG]*AATTCCGG\(.\)\(.\)\(.\)\(.\)[ATCG]*$/\1\2\3\4/p; 0,/CCGGAATT/s/^[ATCG]*\(.\)\(.\)\(.\)\(.\)CCGGAATT[ATCG]*$/\4\3\2\1/p; y/ATCG/TAGC/'

However, the last script y/ATCG/TAGC/, is being ignored
I solved the problem with y/ATCG/TAGC/

Code:

sed -n '0,/AATTCCGG/s/^[ATCG]*AATTCCGG\(.\)\(.\)\(.\)\(.\)[ATCG]*$/\1\2\3\4/p; /AATTCCGG/!y/ATCG/TAGC/; s/^[ATCG]*\(.\)\(.\)\(.\)\(.\)GGCCTTAA[ATCG]*$/\4\3\2\1/p'

But if I add 0,/y/ATCG/TAGC/ to limit the extent of the script to the first occurrence exclusively
I guess I got it:

Code:

sed -n '0,/AATTCCGG/s/^[ATCG]*AATTCCGG\(.\)\(.\)\(.\)\(.\)[ATCG]*$/\1\2\3\4/p; 0,/CCGGAATT/y/ATCG/TAGC/; s/^[ATCG]*\(.\)\(.\)\(.\)\(.\)GGCCTTAA[ATCG]*$/\4\3\2\1/p'