Thread Subject:

word occurence counting (DNA)

Hello,
how can i calculate the frequencies of appearance of the words with K-length walking through a DNA sequence, I want to find the frequency for each path i walk

"ambrosia nightwish" <mess_imen@yahoo.fr> wrote in message <hp0f8k$snf$1@fred.mathworks.com>...
> Hello,
> how can i calculate the frequencies of appearance of the words with K-length walking through a DNA sequence, I want to find the frequency for each path i walk

what have YOU done so far to solve YOUR particular problem...

us

I used FCGR toolbox to calculate the frequency of appearance of all the words with K length , I obtained a final matrix of frequencies.The purpose is to find the frequency of every word reads during the process of calculation.

"ambrosia nightwish" <mess_imen@yahoo.fr> wrote in message <hp51me$adm$1@fred.mathworks.com>...
> I used FCGR toolbox to calculate the frequency of appearance of all the words with K length , I obtained a final matrix of frequencies.The purpose is to find the frequency of every word reads during the process of calculation.

but then - have you not already solved the problem(?)...
otherwise, show a small input/output example...

us

no not yet

"ambrosia nightwish" <mess_imen@yahoo.fr> wrote in message <hp53f7$97n$1@fred.mathworks.com>...
> no not yet

...and the example(?)...

us

in the FCGR toolbox of Jesús Mena-Chalco we have an example

if i have the sequence: AACCGTTAACGT, and i want to find the frequencies of all the dinucleotides (word of 2 letters for example), at the position 1 we read AA so the frequency of appearence is freq=1/12 , at the second position w read AC and freq=1/12, at the eighth position AA appeared fo the second time so freq=2/12 ,the calcul is stopped at the position N-len+1 (N:length of the sequence, len: length of word)

"ambrosia nightwish" <mess_imen@yahoo.fr> wrote in message <hp54tn$2ii$1@fred.mathworks.com>...
> if i have the sequence: AACCGTTAACGT, and i want to find the frequencies of all the dinucleotides (word of 2 letters for example), at the position 1 we read AA so the frequency of appearence is freq=1/12 , at the second position w read AC and freq=1/12, at the eighth position AA appeared fo the second time so freq=2/12 ,the calcul is stopped at the position N-len+1 (N:length of the sequence, len: length of word)
--------------
  Here's an outline of how you might go about it. Let v be a vector of length N with the nucleotide sequence - I am assuming they are represented by four numbers in this discussion - and k be the desired word length.

1) Create a N-k+1 by k matrix, M, containing the successive length-k words. You can use the 'hankel' function for this purpose.

2) Apply [B,m,n] = unique(M,'rows') to M. B will be a table of all the words appearing in the sequence in sorted order.

3) Apply 'histc' to the vector n to obtain the counts of B words in the sequence.

4) From the counts you can obtain the frequencies.

  Can you take it from there?

Roger Stafford

"ambrosia nightwish" <mess_imen@yahoo.fr> wrote in message <hp54tn$2ii$1@fred.mathworks.com>...
> if i have the sequence: AACCGTTAACGT, and i want to find the frequencies of all the dinucleotides (word of 2 letters for example), at the position 1 we read AA so the frequency of appearence is freq=1/12 , at the second position w read AC and freq=1/12, at the eighth position AA appeared fo the second time so freq=2/12 ,the calcul is stopped at the position N-len+1 (N:length of the sequence, len: length of word)

one of the many solutions

% the data
     s='AACCGTTAACGT';
     wl=2;
% the engine
     rpat=sprintf('\\S{%d,%d}',wl,wl);
     t=cell(wl,1);
for i=1:wl
     t{i,1}=regexp(s(i:end),rpat,'match').';
end
     t=cat(1,t{:});
     [tu,ix,ix]=unique(t);
     n=histc(ix,1:max(ix));
     r=[tu,num2cell(n)];
% the result
     disp(s);
     disp(r);
%{
% S =
     AACCGTTAACGT
% R =
     'AA' [2]
     'AC' [2]
     'CC' [1]
     'CG' [2]
     'GT' [2]
     'TA' [1]
     'TT' [1]
%}

us

On Fri, 02 Apr 2010 13:10:31 -0300, ambrosia nightwish
<mess_imen@yahoo.fr> wrote:

> if i have the sequence: AACCGTTAACGT, and i want to find the frequencies
> of all the dinucleotides (word of 2 letters for example), at the
> position 1 we read AA so the frequency of appearence is freq=1/12 , at
> the second position w read AC and freq=1/12, at the eighth position AA
> appeared fo the second time so freq=2/12 ,the calcul is stopped at the
> position N-len+1 (N:length of the sequence, len: length of word)


   s='AACCGTTAACGT';

   wLen=2;

   %associative array, hash, lookup table ...(please see help)
   countMap = containers.Map();

   for indx=1: length(s)-wLen+1

       curWord = s(indx:indx+wLen-1);

       if(isKey(countMap,curWord))
           %we have seen this, increment count
           countMap(curWord)=countMap(curWord)+1;
       else
           countMap(curWord)=1;
       end

   end

   words = countMap.keys;


   frequency = countMap.values;
   %Convert to an array
   frequency = [frequency{:}];

   prob = frequency./sum(frequency)
 

THe problem still exists:The first solution shows the number of the counted words and gives a final result what I want to do is to find the number of appearance of words in every step i walk (increment by 1and word reading by wl), Let us take the same example s='AACCGTTAACGT'
for the words:
AAC: n=1
ACC : n=1
CCG: n=1
CGT: n=1
TTA: n=1
TAA: n=1
AAC: n=2
ACG: n=1
CGT: n=2
AS for the second solution, the containers.Map function dont exist in the matlab version that i have.

Something like this?

s = 'AACCGTTAACGT';
k = 3;

d = double(s);
A = hankel(d(1:end-k+1),d(end-k+1:end));
[u i j] = unique(A,'rows');
b = zeros(length(i),1);
c = zeros(size(j));
for n=1:length(j)
    jn = j(n);
    b(jn) = b(jn)+1;
    c(n) = b(jn);
end

S = char(A)
c

% Bruno

That's working Bruno, thank you all

% Here is an vectorized code (not necessary meant faster)
% http://www.mathworks.com/matlabcentral/fileexchange/24255

s = 'AACCGTTAACGT';
k = 3;

d = double(s);
A = hankel(d(1:end-k+1),d(end-k+1:end));
[u i j] = unique(A,'rows');
[js is]=sort(j);
clear c
c(is) = cell2mat(SplitVec(js,[],@(x) (1:length(x))')) % SplitVec on FEX

% Bruno

"Bruno Luong" <b.luong@fogale.findmycountry> wrote in message <hp9pt1$go1$1@fred.mathworks.com>...
> Something like this?
>
> s = 'AACCGTTAACGT';
> k = 3;
>
> d = double(s);
> A = hankel(d(1:end-k+1),d(end-k+1:end));
> [u i j] = unique(A,'rows');
> b = zeros(length(i),1);
> c = zeros(size(j));
> for n=1:length(j)
> jn = j(n);
> b(jn) = b(jn)+1;
> c(n) = b(jn);
> end
>
> S = char(A)
> c
>
> % Bruno

How to convert S into a vector??

"ambrosia nightwish" <mess_imen@yahoo.fr> wrote in message <hq2q20$aqn$1@fred.mathworks.com>...
> "Bruno Luong" <b.luong@fogale.findmycountry> wrote in message <hp9pt1$go1$1@fred.mathworks.com>...
> > Something like this?
> >
> > s = 'AACCGTTAACGT';
> > k = 3;
> >
> > d = double(s);
> > A = hankel(d(1:end-k+1),d(end-k+1:end));
> > [u i j] = unique(A,'rows');
> > b = zeros(length(i),1);
> > c = zeros(size(j));
> > for n=1:length(j)
> > jn = j(n);
> > b(jn) = b(jn)+1;
> > c(n) = b(jn);
> > end
> >
> > S = char(A)
> > c
> >
> > % Bruno
>
> How to convert S into a vector??

what kind of ...vector... do you mean(?)...

us

ambrosia nightwish wrote:
...

> How to convert S into a vector??

You mean like

S(:)' % ?

--

"Bruno Luong" <b.luong@fogale.findmycountry> wrote in message <hp9pt1$go1$1@fred.mathworks.com>...
> Something like this?
>
> s = 'AACCGTTAACGT';
> k = 3;
>
> d = double(s);
> A = hankel(d(1:end-k+1),d(end-k+1:end));
> [u i j] = unique(A,'rows');
> b = zeros(length(i),1);
> c = zeros(size(j));
> for n=1:length(j)
> jn = j(n);
> b(jn) = b(jn)+1;
> c(n) = b(jn);
> end
>
> S = char(A)
> c
>
> % Bruno

How to convert S into a vector??

"ambrosia nightwish" <mess_imen@yahoo.fr> wrote in message <hq3213$eua$1@fred.mathworks.com>...
> "Bruno Luong" <b.luong@fogale.findmycountry> wrote in message <hp9pt1$go1$1@fred.mathworks.com>...
> > Something like this?
> >
> > s = 'AACCGTTAACGT';
> > k = 3;
> >
> > d = double(s);
> > A = hankel(d(1:end-k+1),d(end-k+1:end));
> > [u i j] = unique(A,'rows');
> > b = zeros(length(i),1);
> > c = zeros(size(j));
> > for n=1:length(j)
> > jn = j(n);
> > b(jn) = b(jn)+1;
> > c(n) = b(jn);
> > end
> >
> > S = char(A)
> > c
> >
> > % Bruno
>
> How to convert S into a vector??

instead of just dumbly repeating your post, you should answer the questions people have asked you...
your performance here in CSSM is not boding well for your future...

us

"ambrosia nightwish" <mess_imen@yahoo.fr> wrote in message <hq3213$eua$1@fred.mathworks.com>...
>
> How to convert S into a vector??

Why you want convert it to"vector" (whatever "vector" is; which nobody seems to know beside you)?

To return the substring at 6th position, just call
substr = S(6,:)

The storage should be appropriate for any further manipulation. No need to convert to vector or vecteur or cell or arrow or flèche or pointeur etc

Bruno

>The solution which I looked for was given by Bruno. Thank you all.
best regards