VarQ: A Tool for the Structural and Functional Analysis of Human Protein Variants

Abstract

Understanding the functional effect of Single Amino acid Substitutions (SAS), derived from the occurrence of single nucleotide variants (SNVs), and their relation to disease development is a major issue in clinical genomics. Despite the existence of several bioinformatic algorithms and servers that predict if a SAS is pathogenic or not, they give little or no information at all on the reasons for pathogenicity prediction and on the actual predicted effect of the SAS on the protein function. Moreover, few actual methods take into account structural information when available for automated analysis. Moreover, many of these algorithms are able to predict an effect that no necessarily translates directly into pathogenicity. VarQ is a bioinformatic pipeline that incorporates structural information for the detailed analysis and prediction of SAS effect on protein function. It is an online tool which uses UniProt id and automatically analyzes known and user provided SAS for their effect on protein activity, folding, aggregation and protein interactions, among others. We show that structural information, when available, can improve the SAS pathogenicity diagnosis and more important explain its causes. We show that VarQ is able to correctly reproduce previous analysis of RASopathies related mutations, saving extensive and time consuming manual curation. VarQ assessment was performed over a set of previously manually curated RASopathies (diseases that affects the RAS/MAPK signaling pathway) related variants, showing its ability to correctly predict the phenotypic outcome and its underlying cause. This resource is available online at http://varq.qb.fcen.uba.ar/. Supporting Information & Tutorials may be found in the webpage of the tool.

Keywords: FoldX; bioinformatics; single amino acid substitutions; single amino acid substitutions classification; variation diagnosis; web server.

FIGURE 1

VarQ general pipeline. Each known structural conformation of the input protein is analyzed independently to aid the user in variant effect interpretation.

FIGURE 2

An Example of VarQ output for the HRAS gene (PDB id 3DDC). On top left, we show the structural features derived from crystallographic data. Target gene and possible interacting genes are shown in green and gray respectively. Structural coverage, PFam family assignment and location of the variant are mapped over the UniProt original sequence.

FIGURE 3

Change in folding energy (ΔΔG) computed by the FoldX software for all mined variants in analyzed RASopathies related proteins. Pathogenic and non-pathogenic variants are shown separately.

FIGURE 4

Receiver Operating Characteristic Curves for the diagnosis of Kiel & Serrano manually curated pathogenic variants that were mapped into structure. Pathogenic mutations are considered true positives in Clinvar (black curve) data set when they are labeled as “pathogenic” or “likely pathogenic.” For VarQ (green curve) true positives are considered for all the labels except “no effect”.