Thanks for clarifying. You still haven't my answered my question about whether or not a character can be part of several matches or if the matches can intersect (see my 'a***a***b***b' example).
Assuming the answer is no to both questions, the simplest way is to use a regular expression: stringa1='{a(abc)(ac)d(cf)}';
stringa2='{(ad)c(bc)(ae)}';
stringa3='{(ef)(ab)(df)cb}';
stringa4='{eg(af)cbc}';
stringa={stringa1; stringa2; stringa3; stringa4};
matchstarts = regexp(stringa, 'a.+?b');
matchcounts = cellfun(@numel, matchstarts)
The regular expression above says match 'a', followed by as little as necessary (the ?) but at least one (the +) characters, followed by a 'b'.
For 'ba', the regular expression would then be 'b.+?a'
Note that if the answer is yes to either question, regular expressions won't work.
