Finding the repeated substrings

I have a DNA sequence that is AAGTCAAGTCAATCG and I split into substrings such as AAGT,AGTC,GTCA,TCAA,CAAG,AAGT and so on. Then I have to find the repeated substirngs and their frequency counts ,that is here AAGT is repeated twice so I want to get AAGT - 2.How is this possible .

[Stephen23]

See Andrei Bobrov's answer for an efficient solution.

[Andrei Bobrov]

Thank you Stephen!

Answer 1

[KSSV]

str = {'AAGT','AGTC','GTCA','TCAA','CAAG','AAGT'} ;
idx = cellfun(@(x) find(strcmp(str, x)==1), unique(str), 'UniformOutput', false) ;

L = cellfun(@length,idx) ;

Ridx = find(L>1) ;

for i = 1:length(Ridx)
    st = str(idx{Ridx}) ;
    fprintf('%s string repeated %d times\n',st{1},length(idx{Ridx}))
end

Answer 2

[Andrei Bobrov]

A = 'AAGTCAAGTCAATCG';
B = hankel(A(1:end-3),A(end-3:end));
[a,~,c] = unique(B,'rows','stable');
out = table(a,accumarray(c,1),'VariableNames',{'DNA','counts'});

[Anthony Tracy]

If it's alright, I had a question about the use of unique. Why not use tabulate? Just curious.

Thanks!

[Image Analyst]

Maybe he didn't know about it - I didn't.

outT = tabulate(B)

out =
  8×2 table
    DNA     counts
    ____    ______
    AAGT      2   
    AGTC      2   
    GTCA      2   
    TCAA      2   
    CAAG      1   
    CAAT      1   
    AATC      1   
    ATCG      1   
outT =
  8×3 cell array
    {'AAGT'}    {[2]}    {[16.6666666666667]}
    {'AGTC'}    {[2]}    {[16.6666666666667]}
    {'GTCA'}    {[2]}    {[16.6666666666667]}
    {'TCAA'}    {[2]}    {[16.6666666666667]}
    {'CAAG'}    {[1]}    {[8.33333333333333]}
    {'CAAT'}    {[1]}    {[8.33333333333333]}
    {'AATC'}    {[1]}    {[8.33333333333333]}
    {'ATCG'}    {[1]}    {[8.33333333333333]}

[Anthony Tracy]

yeah that's fair. I was just curious since I was just looking at both and wondering why I may want to use one over the other. Seems mainly like if I want a table or cell.

Thanks!

[Stephen23]

tabulate requires the Statistics and Machine Learning Toolbox, which not everyone has.

[Ivan Savelyev]

Hi.

I have a question. Some time i have a ladder-like results (nested sequences) like this :

AAAAAAAAA which will be calculated (with frame size 3 as) as 6 AAAA sequences, wich is not correct in some cases ( it is also about ATATATA type of sequences). Is there a solution or algorithms to filter nested repeats ?

Thanx a lot.

Answer 3

[Steven Lord]

For the original question you could convert the char data into a categorical array and call histcounts.

>> C = categorical({'AAGT','AGTC','GTCA','TCAA','CAAG','AAGT'})
C = 
  1×6 categorical array
     AAGT      AGTC      GTCA      TCAA      CAAG      AAGT 

>> [counts, uniquevalues] = histcounts(C)
counts =
     2     1     1     1     1

uniquevalues =
  1×5 cell array
    {'AAGT'}    {'AGTC'}    {'CAAG'}    {'GTCA'}    {'TCAA'}