An old PAM35 exists , having a value of 3,5 for identities (elements of the main diagonal) . However, this matrix is too crude, making R and H equivalent to F and Y for example. That's why I looking for a BLOSUM62 having also the diagonal elements set to a same value. The NCBI has a lot of BLOSUM matrices, none of them fitting my willings.
Maybe someone can tell me how to compute such a matrix ?
The numbers in the matrices are related to the probability of seeing a certain pair of amino acids aligned in sequence alignments (see https://en.wikipedia.org/wiki/BLOSUM). While you can easily scale the matrices (via the value of lambda), you can't set the entire diagonal to one without changing the underlying assumptions behind the matrix (e.g. because it is much more likely that an alignment of two leucines is a result of chance than it is for the much rarer cysteines or tryptophans - and so you HAVE TO score them differently).
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Bioinformatics