how to replace characters into digits

i have to replace the each characters using the following digits s=ACGT

i have to replace as

        'A' then 11

        'C' then 00

        'G' then 01

        'T' then 10

[Jan]

You write 11 without surrounding quotes. This isn't an accident, correctly? What do you want as output? How long is the input?

Answer 1

[Azzi Abdelmalek]

clear 
s='ACGT'
e=['11';'00';'01';'10']
in='AGCTAG'      % Your initial data
out=in
for k=1:numel(s)
 out=regexprep(out,s(k),e(k,:))
end

[Sharen H]

Thanks a lot....

[Jan]

STRREP is less powerful, but much faster than REGEXPREP.

[Cedric]

STRREP is probably the best solution. It is a little slower than my solution, but more memory friendly.

Answer 2

[Jan]

And a lookup table:

seqIn = 'ACGTTGCA'
table = repmat('0', 2, 255);
table(1, 'AT') = '1';
table(2, 'AG') = '1';
result = reshape(table(:, seqIn), 1, []);

Does this work? I do not have access to Matlab currently.

[Azzi Abdelmalek]

Yes It works

Azzi (regexprep),Azzi (strrep),Cedric,Jan execution time in this order

125000 character

Elapsed time is 14.543628 seconds.
Elapsed time is 0.007702 seconds.
Elapsed time is 0.005915 seconds.
Elapsed time is 0.006354 seconds.

[Cedric]

Actually STRREP still wins up to me. The solution based on ISMEMBER is profiled between REGEXP and the two solutions based on array indexing. But the latter require much more memory during the indexing operation than STRREP. You start seeing that with ~1e8 chars. On my laptop with 8GB RAM for example, only STRREP can treat more than 2e8 chars without swapping (or being killed by Process Lasso).

[Jan]

@Azzi: Is this a typo?! Your function needs 14 secs with REGEXPREP and 0.007 secs with STRREP? Then my minor suggestion caused a speedup of a factor 1900? Wow, this would be the most efficient suggestion I ever gave. And it would be a strong hint to warn for the low efficiency of regexprep in this forum.

Answer 3

[Cedric]

If you need to process long sequences, you might want to optimize a little the efficiency.. a MEX-based solution would be most efficient I guess, but here is one way you could go using basic MATLAB only..

If you want to replace character 'A' with a numeric array (1,1) and so on, you can do:

 aa  = 'ACGT' ;
 seq = 'AAGCTCAGGTTCA' ;

 rep = zeros(2, max(aa), 'uint8') ;
 rep(:,aa) = [1 0 0 1; 1 0 1 0] ;

 result = reshape(rep(:,seq), 1, []) ;

This outputs the numeric array:

 result =
   1  1  1  1  0  1  0  0  1  0  0  0  1  1  0  1  0  1  1  0  1  0  0  0  1  1

If you want to replace character 'A' with characters '11' and so on, you can do:

 aa  = 'ACGT' ;
 seq = 'AAGCTCAGGTTCA' ;

 rep = zeros(2, max(aa), 'uint8') ;
 rep(:,aa) = ['11'; '00'; '01'; '10'].' ;

 result = char(reshape(rep(:,seq), 1, [])) ;

This outputs the string '11110100100011010110100011'.

EDIT: note that there are slightly different ways of doing this a little slower but with a more memory-friendly approach.

Cheers,

Cedric

Answer 4

[Jos (10584)]

Here is a simpler approach than looping over REGEXPREP or STRREP:

seqIn = 'ACGTTGCA' % input sequence

letters = 'ACGT' ;
symb = {'11','00','10','01'} ; % stored as a cell array of strings!
[tf,idx] = ismember(seqIn,letters) ;
seqOut = [symb{idx}]