I am not sure what all you want to keep, but your regex seems a little extreme for the example you show. Here is how I would solve it. I give it a little bit of formatting that I hope doesn't confuse the issue.

while (<DATA>) {
        # the regex remembers for one or more 
        # alphanumeric characters \w+
        # followed by an equal sign and double quote
        # then remembers one or more digit or comma [\d\,]+
        # follwed by a double quote.
    if (/(\w+)\=\"([\d\,]+)\"/) {
        

        # The print statement adds an equal amount
        # of padding between the first entry and the
        # second by subtracting the length of $1 by
        # twenty and then adding that many spaces
        # to the output.
        # replace ', " " x (20 - length $1)'
        # with ', "\t"' if you want a tab instead
        print ucfirst($1) , " " x (20 - length $1) , $2 , "\n"; 
    }
}


__DATA__
FEATURE              /group=" "
                       /translation="MMSKLGVLLT ICLLLFPLTA VQLDGDQPAD 
+LPALRTQDIA  
                       TDHSPWFDPV KRCCSRYCYI CIPCCPN"  
                       /disulfide="9,19"
                       /disulfide="13,24"
                       /hydroxylation="10"  
                       /hydroxylation="11"
                       /post_trans_mod="
[download]

update: Added code comments