Predicted Effects of hTP Variants.

*PolyPhen-2 provides two results based on HumVar (13,032 human disease causing mutations from UniProt and 8,946 human nonsynonymous single-nucleotide polymorphisms (nsSNPs) and HumDiv [29]. These prediction algorithms return a scaled probability from 0 (neutral) to >1 (damaging).
†The Grantham score [32] represents the original amino acid substitution scoring method, presenting a scale increasing from zero to indicate the degree of side-chain physicochemical difference between the original and mutant amino acid. Single digit scores are considered conservative, with severe mutations ranging 60 or higher.
‡The SIFT (Sorting Intolerant From Tolerant) algorithm (http://blocks.fhcrc.org/sift/SIFT.html) [26] infers functional importance from sequence homology. Based on a PSI-BLAST search alignment, SIFT returns a scaled probability matrix for the likelihood of a protein to tolerate each of the twenty amino acids at each position in the protein. Output values for each amino acid change tolerance from SIFT ranges from 0 (damaging) to 1 (neutral).
§ProPhylER (http://www.prophyler.org) [31] combines the physicochemical properties of amino acid side chains and the observed evolutionary variation of those properties to infer deleterious substitutions.
║The module of the SNPs3D resource (http://www.SNPs3D.org) [30] which focuses on predicting SNP influence on protein function, separately evaluates protein structural stability analysis and sequence conservation using a support vector machine approach trained on monogenic disease. Negative values are considered deleterious, positive values considered neutral, and values beyond ±0.5 indicate the degree of confidence.
#Position specific scoring matrix returns sequence alignment-based probability, with lower scores for damaging variants.