What we think Darwin does best

1. Local Alignments at Best PAM
   This is a simple alignment between two amino acid sequences, where we adjust the scoring matrix (by adjusting the PAM value, or evolutionary distance between the sequences). This gives a maximum likelihood alignment. Very distant sequences are typically aligned better with this technique.
2. Multiple Sequence Alignments / Probabilistic Ancestral Sequences
   Darwin can compute a MSA together with PAS. A PAS is a probability profile of the possible amino acids in each position of the root of the tree. The MSA derived in this way (and with the additional heuristics by Chantal Korostensky) is usually much better than the ones produced by other methods.
3. PAM distance between sequences
   Darwin can compute, statistically sound, distances between sequences, including their variance. This is also used to build phylogenetic trees. Darwin's algorithm for building trees based on distance matrices (optionally with variances) is very good.
4. Mass profile
   We call mass profiling the searching of a protein database based on the digestion of a protein and its reading with a mass spectrometer. We do this for protein or for RNA/DNA databases, with the appropriate conversions. A dynamic programming search of a protein database given a fragment of a protein and its partial weights is also available. (i.e. all/some of the weights of subfragments of a fragment).
5. Nucleotide-Peptide alignment
   Darwin does a nucleotide-peptide alignment which includes the genetic code table in the scoring function. This has the advantage of allowing the detection of deletions in the RNA/DNA that cause frame shifts and the clear detection of introns.
6. Least Squares, Best Basis
   This is a purely statistical facility. Darwin computes the best subset of variables that will fit certain data. This is the full solution of the problem sometimes called "stepwise regression" in statistics. It is usually a very useful tool when given a large number of observed variables and when we want to determine which are the ones best explaining certain phenomenon.