Generalise the alignment score to one peptide against a set of germlines

[douweschulte]

If you align one peptide against multiple germlines it is nice to have a way of determining which alignments are interesting. If this peptide can reasonably be assumed to come from one germline, it shares too much with a subset, or it does not match any at all.

Older thoughts: 1 score per germline for how likely to be enriched, taking into account that germlines can be distanced heterogeneously, that peptides can match incorrectly and the peptides can be plain wrong.

When this abs score is normalised by the number of peptides it gives a feeling on how well peptides support this specific germline. Alternatively when normalised on the maximal absolute germline score it gives a feeling on the absolute number of support (and with that the number of peptides).