CDPred (Conserved Domain-based Prediction) is a computational algorithm that is designed to theoretically calculate the effect of substituting an amino acid relative to the reference sequence within functional modules – the protein domains. Effectively, CDPred computes the deviation (as reported by delta-score) from the expected in a position-specific manner, based on domain annotations in the NCBI Conserved Domain Database (CDD). So this method was designed to analyze the missense variation and mutations in humans (though can be used for any organism) in the light of conservation within functional units of the protein. Lower the value of delta-score (more negative the score), more severe or deleterious the predicted effect on the function of the protein. We set 0 to -3 as mild / neutral effects, -4 to -6 as medium effects, and -7 and lower as severe effects. A severity score of -30 is assigned to termination or stop changes, as they would lead to a truncated protein product. It must be noted that positive scores above 3 (high positive-scores) may also be potentially damaging, but a strongly positive score generally results from a situation where the normal human amino acid at a position is different from many or all other aligned species over a domain, and that the variant allele is closer to the ancestral form. For example, if an individual had the ancestral form of the FOXP2 gene, that individual would potentially exhibit loss of speech, however, the delta score for the amino acid positions that are unique to all other humans would be positive because that individual more closely resembles the ancestral form.
:: MORE INFORMATION
Johnston, J.J., Teer, J.K., Cherukuri, P.F., Hansen, N.F., Loftus, S.K., NIH Intramural Sequencing Center, Chong, K., Mullikin, J.C., Biesecker, L.G. 2010.
Massively parallel sequencing of exons on the X chromosome identify RBM10 as the gene that causes a syndromic form of cleft palate.
Am J Hum Genet. 2010 May 14;86(5):743-8.