Conformal predictors in Identifying Protein sequences

To assess the viability of conformal prediction in understanding the accuracy of tools in identifying protein sequences with particular protein families. An important task in Bioinformatics is identifying if a protein sequence is part of a particular protein family. This is typically carried out using tools such as Hidden Markox Models (HMM's). The difficulty with such models is that while they produce a probabilistic likelihood of a hit, the probabilities are based on a number of modelled parameters. Typically, such probabilities generically generate incredibly small p-values and yet some suspician remains that these methods are not robust as the results suggest.

A simple implementation of conformal prediction.
A framework to evaluate the credibility and confidence of individual HMM's from PFAM.