I have large data file, like this:
AM_fep=1;AF=8.236e-06;AN=121412…
AM_fep =1; AN=121412;AF=0.000265…
AM_fep =2;AF=1.647e-05;AN=121412…
I require to grep only AF= field with its numeric value.
I have used grep -o -E 'AF=[0-9]*' but it gives the first digit of the value as:
AF=8
AF=0
AF=1
Given your corpus, you can field-split on semicolons and then loop over each field on the line looking for AF as a match. For example:
$ awk -F\; '{for (i=1; i<=NF; i++) if ($i ~ /AF/) print $i}' /tmp/corpus
AF=8.236e-06
AF=0.000265…
AF=1.647e-05
Note that the second example correctly matched the … ellipsis character contained in your posted corpus. Your real data probably doesn't have them, but if it does you can use the match() or sub() string functions to exclude it. As an example:
$ awk -F\; '{for (i=1; i<=NF; i++) if ($i ~ /AF/) {sub(/…/, "", $i); print $i}}' \
/tmp/corpus
AF=8.236e-06
AF=0.000265
AF=1.647e-05
This requires using grep's -o option which only prints the part of the line that matched the pattern.
We can match this with
grep -o 'AF=[^;]*' datafile
The pattern matches the literal string AF= followed by any number of non-semicolon characters. Against your file, it produces
AF=8.236e-06
AF=0.000265…
AF=1.647e-05
Using egrep (or equivalently grep -E) we can require that there be at least one character after the equal sign with
egrep -o 'AF=[^;]+' datafile
The reason that your attempted pattern is not working is that it matches AF= followed by any number of digits. Thus it can not pick up the decimal point (or the exponent symbol and negative sign).
