**50**wrote:

Hi

I want to do global sequence aligment. Referring to the wikipedia article of Needleman-Wunsch for global sequence alignment. I could see the following similarity matrix

```
A G C T
A 10 -1 -3 -4
G -1 7 -5 -3
C -3 -5 9 0
T -4 -3 0 8
```

Instead of having these values I want to deal with the probability. So all the values will not be integers but real values.

In all of the matrices I had seen e.g BLOSUM50, all have the integer values. My question is that if the similarity matrix will be filled with real values. Would it work the same way.

Another thing is that, is there any special procedure to construct the similarity matrix or it could be any distance measure.

Thanks a lot.

Wouldn't you rather want to use a hidden markov model when working with probabilities?

2.5k