I have ~17k amino acid sequences in FASTA format in a single file. Using following command of Clustal Omega on Linux system, I created the distance matrix;
clustalo -i filename.faa --distmat-out=filename.faa.mat --full
It created a matrix of 17k fields x 17k records. The matrix is displayed in following format;
A B C D E ....
A 0.000 0.136 0.227 0.476 0.864
B 0.136 0.000 0.318 0.571 0.864
C 0.227 0.318 0.000 0.238 0.773
D 0.476 0.571 0.238 0.000 0.857
E 0.864 0.864 0.773 0.857 0.000
.
.
Manipulating this type of data is difficult for me as many values are repeating. I want the distance matrix to be drawn in following format;
A B 0.136
A C 0.227
A D 0.476
A E 0.864
B C 0.318
B D 0.571
B E 0.864
C D 0.238
C E 0.773
D E 0.857
.
.
Dealing with this data containing info of 17k sequences will be relatively easy for me.
Can anybody help me how to convert the format of distance matrix.
I'm not sure what programming language you're writing in but a trivial way of doing this would be to (bit of pseudo-code):
For each horizontal line (i):
1) Record the numbers at and after i + 1 and assign the column names of (i + 1):N
2) Append the i'th row name to all of the names in 1).
Hi Muhammad
I believe it is easy to convert the matrix to the vector format you prefer.
if A is a matrix with dimensions m x n, you can create a vector by issuing the command (in MATLAB) : B=reshape(A,m*n,1). B will be the vector you are interested in. It will only contain the numbers of the matrix A - not the column and row assignments, but you can deduce this without too much trouble from the location in the vector.
Alternatively if you do not have access to MATLAB you can construct your own 'reshape' function :
count=0;
for i=1:m
for j=1:n
count=count+1
B(count)=A(i,j);
end
end
where A is your matrix and B is the vector you are interested in. Here you also see that count maps uniquely onto the i,j matrix location.
Hope this may help
Regards
Steffen
Dear Dave Lee and Steffen Petersen,
Thanks for your answers, but I do not have experience on MATLAB. I am good in Shell scripting, and beginner of Perl and R language. Any solution in these would be beneficial for me.
Hey Muhammad,
My pseudo-code was written specifically with R in mind (I'm predominantly an R-guy). I however wouldn't recommend reading in a 17k^2 matrix as I don't think R will like it!
If you want to do it in R, I would recommend opening a connection to the file using file(). Then in a for loop, readLines() the file but set the number of lines to read to 1 (so it reads one-by-one). To get the numbers out, I'm guessing they're delimited e.g. with a tab so strsplit() them and record the numbers after i = i (otherwise you get redundancy from the lower diagonal of the matrix which you want to avoid). Finally, paste the i'th rowname with all the colnames that are i + 1 with paste(with sep = "-").
Sorry for the rambling but this is how I would do it in R.
- Badges
- Science topic
- Similar topics
- Bioinformatics
- Bioinformatics and Computational Biology