Molecular damage inFabry disease: Characterization and prediction of alpha‐galactosidaseApathological mutations

Abstract

Loss-of-function mutations of the enzyme alpha-galactosidase A (GLA) causes Fabry disease (FD), that is a rare and potentially fatal disease. Identification of these pathological mutations by sequencing is important because it allows an early treatment of the disease. However, before taking any treatment decision, if the mutation identified is unknown, we first need to establish if it is pathological or not. General bioinformatic tools (PolyPhen-2, SIFT, Condel, etc.) can be used for this purpose, but their performance is still limited. Here we present a new tool, specifically derived for the assessment of GLA mutations. We first compared mutations of this enzyme known to cause FD with neutral sequence variants, using several structure and sequence properties. Then, we used these properties to develop a family of prediction methods adapted to different quality requirements. Trained and tested on a set of known Fabry mutations, our methods have a performance (Matthews correlation: 0.56-0.72) comparable or better than that of the more complex method, Polyphen-2 (Matthews correlation: 0.61), and better than those of SIFT (Matthews correl.: 0.54) and Condel (Matthews correl.: 0.51). This result is validated in an independent set of 65 pathological mutations, for which our method displayed the best success rate (91.0%, 87.7%, and 73.8%, for our method, PolyPhen-2 and SIFT, respectively). These data confirmed that our specific approach can effectively contribute to the identification of pathological mutations in GLA, and therefore enhance the use of sequence information in the identification of undiagnosed Fabry patients.

The Complementarity Between Protein-Specific and General Pathogenicity Predictors for Amino Acid Substitutions

“…In this approach, we cannot exclude the possibility that some of the variants retrieved happen at MSA loci involved in a network of correlations with other loci, a situation indicative of a functional role [Kowarsch et al., ; Hecht et al., ]. However, we expect this number to be small, because neutral variants tend to happen at locations where sequence variability is higher [Riera et al., ] and these positions generally have undetectable or numerically low correlations [Fodor and Aldrich, ]. For this reason, and because the presence of these variants would affect both PSP and GM, we do not expect a substantial impact on our main results.…”

“…The latter, which is the focus of this work, has been explored in a small number of studies, where specific predictors are obtained by training them only with variants from the protein(s) of interest. For example, our recent Fabry‐specific predictor [Riera et al., ] was trained using a set of 332 pathogenic and 48 neutral variants from alpha‐galactosidase A; the initial version of the predictor for the mismatch repair genes was trained using a set of 168 functionally tested variants (80 pathogenic and 88 neutral) from four genes [Ali et al., ]; Martin et al. () conducted their exhaustive structure analysis in a collection of 882 p53 variants; Izarzugaza et al.…”

The Large Phenotypic Spectrum of Fabry Disease Requires Graduated Diagnosis and Personalized Therapy: A Meta-Analysis Can Help to Differentiate Missense Mutations

“…Other in silico approaches based on the structural features of AGAL, some from our group [64], have been attempted [65,66], to predict the severity of FD genotypes. We believe that data obtained in vitro should always be preferred whenever available.…”

BRCA1 ‐ and BRCA2 ‐specific in silico tools for variant interpretation in the CAGI 5 ENIGMA challenge

“…We used two different MSA, psMSA, and oMSA, which resulted in two versions of the NN predictor. psMSA were obtained using the same protocol utilized for the protein‐specific predictors (Riera et al, , ) which, briefly, consists of two steps: (a) Recovery of BRCA1/2 homologs using a query search of UniRef100; (b) elimination of remote homologs (<40% sequence identity); alignment of the remaining sequences with muscle (Edgar, ). The resulting MSA is available on demand from the authors.…”

“…We used two different MSA, psMSA, and oMSA, which resulted in two versions of the NN predictor. psMSA were obtained using the same protocol utilized for the protein-specific predictors (Riera et al, 2015(Riera et al, , 2016 which, briefly, consists of two steps: (a) Recovery of BRCA1/2 homologs using a query search of UniRef100; (b) elimination of remote homologs (<40% sequence identity); alignment of the remaining sequences with muscle (Edgar, 2004 Three features, each measuring the difference between native and mutant amino acids for a single physicochemical property: van der Waals volume (Bondi, 1964), hydrophobicity scale (estimated from water/octanol transfer free energy measurements) (Fauchere & Pliska, 1983), and the element of the Blosum62 matrix (Henikoff & Henikoff, 1992) corresponding to the amino acid replacement.…”