A new method for predicting the functional impact of protein mutations is introduced, based on evolutionary conservation patterns. The method uses combinatorial entropy formalism to analyze aligned homologous sequences from different species, and calculates a functional impact score (FIS) for amino acid changes. The FIS performs well in distinguishing between disease-associated variants and common polymorphisms, with an area under the receiver operating characteristic curve of ~0.86. In cancer, the method effectively assigns higher scores to mutations more likely to be functional ('drivers'). A list of 1000 top human cancer genes frequently mutated in one or more cancer types is provided, as well as 1000 candidate cancer genes with rare but likely functional mutations. The method also estimates that at least 5% of cancer-relevant mutations involve a 'switch of function', rather than simply loss or gain of function. The method uses evolutionary conservation patterns to predict the functional impact of mutations. It calculates the entropy of residue distribution in an alignment column and estimates the mutation impact based on the difference in entropy caused by the mutation. The method also considers subfamily conservation patterns, using a combinatorial entropy approach to determine specificity residues, which are residues that differ between subfamilies and are likely to be functionally important. The method combines conservation and specificity scores to provide a more accurate prediction of functional impact. The method was validated on a large set of disease-associated and polymorphic variants, and showed good performance in distinguishing between these two types of variants. The method was also tested on experimental data for TP53 mutations, and showed that the FIS score is correlated with the transcriptional activity of mutated TP53. The method was further tested on the COSMIC database, where it showed that recurrent mutations and mutations in frequently mutated genes are more likely to be functional. The method also identified a set of mutations that involve a 'switch of function', which is a new type of functional impact. The method provides a way to prioritize mutations for experimental validation, by ranking mutations based on their predicted functional impact. The method is based on evolutionary conservation patterns and uses a combination of conservation and specificity scores to predict the functional impact of mutations. The method has been validated on a large set of disease-associated and polymorphic variants, and shows good performance in distinguishing between these two types of variants. The method is also useful for identifying mutations that may have a 'switch of function', which is a new type of functional impact. The method provides a way to prioritize mutations for experimental validation, by ranking mutations based on their predicted functional impact.A new method for predicting the functional impact of protein mutations is introduced, based on evolutionary conservation patterns. The method uses combinatorial entropy formalism to analyze aligned homologous sequences from different species, and calculates a functional impact score (FIS) for amino acid changes. The FIS performs well in distinguishing between disease-associated variants and common polymorphisms, with an area under the receiver operating characteristic curve of ~0.86. In cancer, the method effectively assigns higher scores to mutations more likely to be functional ('drivers'). A list of 1000 top human cancer genes frequently mutated in one or more cancer types is provided, as well as 1000 candidate cancer genes with rare but likely functional mutations. The method also estimates that at least 5% of cancer-relevant mutations involve a 'switch of function', rather than simply loss or gain of function. The method uses evolutionary conservation patterns to predict the functional impact of mutations. It calculates the entropy of residue distribution in an alignment column and estimates the mutation impact based on the difference in entropy caused by the mutation. The method also considers subfamily conservation patterns, using a combinatorial entropy approach to determine specificity residues, which are residues that differ between subfamilies and are likely to be functionally important. The method combines conservation and specificity scores to provide a more accurate prediction of functional impact. The method was validated on a large set of disease-associated and polymorphic variants, and showed good performance in distinguishing between these two types of variants. The method was also tested on experimental data for TP53 mutations, and showed that the FIS score is correlated with the transcriptional activity of mutated TP53. The method was further tested on the COSMIC database, where it showed that recurrent mutations and mutations in frequently mutated genes are more likely to be functional. The method also identified a set of mutations that involve a 'switch of function', which is a new type of functional impact. The method provides a way to prioritize mutations for experimental validation, by ranking mutations based on their predicted functional impact. The method is based on evolutionary conservation patterns and uses a combination of conservation and specificity scores to predict the functional impact of mutations. The method has been validated on a large set of disease-associated and polymorphic variants, and shows good performance in distinguishing between these two types of variants. The method is also useful for identifying mutations that may have a 'switch of function', which is a new type of functional impact. The method provides a way to prioritize mutations for experimental validation, by ranking mutations based on their predicted functional impact.