How should I interpret the distribution of C(atm) scores mapped onto the protein surface?
While it is difficult to correlate the absolute value of a C(atm) score to the functional importance of its corresponding atom or residue, mapped C(atm) scores are a useful tool for rapidly identifying potential functional sites by eye and for detecting problems with the input data. Indeed, large low-scoring regions may be indicative of poor conservation, but may equally well be caused by incomplete coverage of the query protein by its homologues. Conversely, high C(atm) scores present over the entire molecule tend to reflect a very low level of diversity in the sequence alignment, which may later manifest itself through decreased prediction accuracy.
