SDM: a server for predicting effects of mutations on protein stability.

Pandurangan AP, Ochoa-Montaño B, Ascher DB, Blundell TL. · 2017

Read SDM: a server for predicting effects of mutations on protein stability. by Pandurangan AP, Ochoa-Montaño B, Ascher DB, Blundell TL. (2017) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc5793720.txt.

Introduction

Pandurangan and co-authors report an updated web server, SDM, that predicts how mutations change protein stability. Such forecasts help sort which genetic variants truly harm proteins. The team rebuilt its knowledge tables from a far larger set of known structures. New measures of packing and burial were added.

The server was tested on 2,690 point mutations across 132 protein structures. It matched reverse mutation checks well, beating machine learning rivals in their tests. A user gives a structure file and a point change. The server returns a score for the stability shift.

Catalog record

CREATORPandurangan AP, Ochoa-Montaño B, Ascher DB, Blundell TL.
YEAR2017
COLLECTIONPAPERS
FORMATTXT
FILE SIZE39 KB
LICENSECC BY 4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/articles/PMC5793720

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

SDM: a server for predicting effects of mutations on protein stability By Arun Prasad Pandurangan, Bernardo Ochoa-Montaño, David B. Ascher, Tom L. Blundell Published in: Nucleic Acids Research DOI: 10.1093/nar/gkx439 ABSTRACT Abstract Here, we report a webserver for the improved SDM, used for predicting the effects of mutations on protein stability. As a pioneering knowledge-based approach, SDM has been highlighted as the most appropriate method to use in combination with many other approaches. We have updated the environment-specific amino-acid substitution tables based on the current expanded PDB (a 5-fold increase in information), and introduced new residue-conformation and interaction parameters, including packing density and residue depth. The updated server has been extensively tested using a benchmark containing 2690 point mutations from 132 different protein structures. The revised method correlates well against the hypothetical reverse mutations, better than comparable methods built using machine-learning approaches, highlighting the strength of our knowledge-based approach for identifying stabilising mutations. Given a PDB file (a Protein Data Bank file format containing the 3D coordinates of the protein atoms), and a point mutation, the server calculates the stability difference score between the wildtype and mutant protein. The server is available at http://structure.bioc.cam.ac.uk/sdm2 INTRODUCTION Recent developments in next-generation sequencing methods have provided a wealth of information on the genetic mutations present in different organisms. In humans, >100 000 genetic variants have been statistically associated with disease conditions (1). The big challenge is to identify and characterize those genetic mutations that have functional consequences. Of particular interest are missense mutations that can disrupt functions of proteins by modulating their stability as well as affecting interactions with other biological molecules. Hence, predicting the impacts of mutations on protein stability and interactions is fundamental to the understanding of various biological processes, including disease and drug resistance (2). While experimental techniques to measure changes in stability between wild-type and mutant proteins are the most accurate, they are time consuming and costly (3). There is a strong need for the development of computational techniques to predict the impacts of mutations on protein stability in order to support the rapid and routine analysis of sequencing data necessary for personalized medicine (4). Methods to predict the impacts of mutations can be broadly classified into sequence- and structure-based approaches. Various sequence-based methods using a broad range of methods have been reported, including support vector machine (INPS) (5), neural networks (6) and decision trees (iPTREE-STAB and MuStab) (7,8). Structure-based methods use either machine learning techniques (9–16) or potential-energy-based approaches to predict the impacts of mutations (17–19). Multi-agent prediction systems, based on statistical scoring functions and machine-learning approaches, have also been reported (20). Various other predictive methods have recently been reviewed elsewhere (21). The development and validation of various computation methods is also supported by databases documenting experimental thermodynamic parameters, including the change in free energy between the wild-type and mutant protein (22,23). Here, we report our updated knowledge-based approach SDM (24,25) and its webserver SDM2 for predicting the effects of mutations on protein stability. SDM pioneered the use of conformationally constrained environment-specific substitution tables (ESSTs) to calculate the change in thermal stability between the wildtype and mutant protein (24–27). SDM has also been successfully used in combination with machine learning techniques to predict better the impact of mutations on protein stability (15). Unlike machine learning methods, SDM predictions do not rely on a number of features for training and do not suffer from the issue of overfitting. SDM2 uses newly recalculated environment-specific substitution tables (ESSTs) for the purpose of calculating the stability difference score between the wildtype and mutant protein structures. New structural parameters, based on residue packing density, have been introduced into the calculation of ESSTs. The newly updated ESSTs were derived from a large set of protein-family sequence and structure alignments, reflecting the current state of fold-classification databases. Below we describe the updated method, webserver and results of the validation process for three different benchmark datasets. MATERIALS AND METHODS Environmental-specific substitution tables SDM uses a set of conformationally-constrained ESSTs to calculate the difference in stability between the wildtype and mutant protein structure (26,27). In SDM2, the updated ESSTs were derived from 2054 protein family sequence and structure alignments from the TOCCATA database, originally developed to serve as a resource for template identification in homology modelling (28) (Ochoa-Montaño B, and Blundell TL, manuscript in preparation), consisting of 12 038 structures. The TOCCATA database incorporates all domains from SCOP 1.75A and CATH 3.5, forming a consensus ‘profile’ whenever the domains of a SCOP family can be reasonably matched to a CATH superfamily, otherwise keeping them in their respective categories. For the calculation of ESSTs we took representative crystal structures (better than 2.5 Å resolution) within each family by following the sequence-clustering procedure using Cd-hit (29), as previously described (30). The program ULLA was used for the purpose of calculating ESSTs (31). ESSTs take the form of probability tables giving details about the amino acid residue conservation and substitution to any other residue occurring in a well-defined local structural environment. They have been shown to capture distinct substitution patterns, specific to a given local structural environment (32). Functional residues, defined as those involved in catalytic site, ligand binding and protein-protein interactions, were identified using CREDO (33) and masked from substitution counts. Previously, in SDM the ESSTs were derived from HOMSTRAD (34) using 371 protein family sequence alignments consisting of 1357 structures. In SDM2, the ESSTs derived from TOCCATA represent a 6- and 9-fold increase in the number of protein families and structures respectively. In this paper, we used two further structural parameters based on residue-occluded packing density (OSP) (35,36) and residue depth (37,38) as alternatives to the relative sidechain solvent accessibility (RSA) (39) parameter used in the calculation of ESSTs. The occluded surface for a given residue represents the molecular surface of the surrounding non-bonded atoms found within 2.8 Å (33,34). The OSP of a residue is calculated as a function of occluded surface area and average unit normal distances between the molecular surfaces of the atoms in a given residue and the neighbouring van der Waals surfaces. The depth of a residue is defined as the average distance of all atom depths found in the residue from the nearest surface water molecule (37,38). RSA can help identify whether a residue is solvent accessible or inaccessible. We have previously proposed that OSP and residue depth could be used to classify better the environment of the interior of the protein (40). OSP and residue depth have been shown to be important in protein structure and stability analysis (37,41,42). Previous studies have shown that the free energy difference between wild-type and mutant proteins (ΔΔG) correlates better with changes in packing parameters such as occluded surface or residue depth than with the change in accessible surface area upon folding (43). Our analyses of the distribution of OSP, depth and RSA based on the TOCCATA dataset suggest that both OSP and depth could be useful structural parameters in defining ESSTs (see Supplementary material, Supplementary Figures S1 and S2). Our recent analysis of ESSTs using the TOCCATA database indicate that the residue conservation progressively increases with residue depth and packing density and could serve as a good indicator for the classification of disease and

Read the complete text