MPRAP: An accessibility predictor for a-helical transmem-brane proteins that performs well inside and outside the membrane

Abstract

BackgroundIn water-soluble proteins it is energetically favorable to bury hydrophobic residues and to expose polar and charged residues. In contrast to water soluble proteins, transmembrane proteins face three distinct environments; a hydrophobic lipid environment inside the membrane, a hydrophilic water environment outside the membrane and an interface region rich in phospholipid head-groups. Therefore, it is energetically favorable for transmembrane proteins to expose different types of residues in the different regions.ResultsInvestigations of a set of structurally determined transmembrane proteins showed that the composition of solvent exposed residues differs significantly inside and outside the membrane. In contrast, residues buried within the interior of a protein show a much smaller difference. However, in all regions exposed residues are less conserved than buried residues. Further, we found that current state-of-the-art predictors for surface area are optimized for one of the regions and perform badly in the other regions. To circumvent this limitation we developed a new predictor, MPRAP, that performs well in all regions. In addition, MPRAP performs better on complete membrane proteins than a combination of specialized predictors and acceptably on water-soluble proteins. A web-server of MPRAP is available at http://mprap.cbr.su.se/ConclusionBy including complete a-helical transmembrane proteins in the training MPRAP is able to predict surface accessibility accurately both inside and outside the membrane. This predictor can aid in the prediction of 3D-structure, and in the identification of erroneous protein structures.

A sequence‐based computational model for the prediction of the solvent accessible surface area for α‐helix and β‐barrel transmembrane residues

“…MPRAP could achieve the prediction correlation coefficient of 0.58. [13] This predictive ability still needs to be improved because MPRAP was developed nondiscriminately for the residues inside and outside the membrane. The methods developed by Yuan et al and MPRAP method are the main methods currently available for predicting the ASA value of TM residues.…”

“…These 78 a-helix membrane protein chains was the same as that used in MPRAP method, [13] thus can be used to directly compare the performance of different ASA prediction method. The 14 b-barrel membrane proteins used in Yuan's method and 10 b-barrel membrane proteins published in 2008 or later were included in this study.…”

“…Illergård also reported a SVM model (named as MPRAP) for the prediction of ASA for the TM residues of ahelical membrane proteins. [13] Since all these methods applied SVM to develop prediction models, here we also selected SVM as the modeling method and compared its performance with that from the random forest algorithm (RF). Both algorithms were utilized to the same dataset with the same 40 conservation score and their performance was assessed by cross-validation.…”

“…By using probes of 1.4, 2.0 Å or a combination of these, Illergård chose a data set of 78 a-TM protein chains to evalute MPRAP and Yuan method. [13] Since it is often difficult to directly compare performance values of different prediction methods reported in different studies, we utilized this data set to evaluate the performance of our model and compare its performance with the reported works. C. Wang et al…”

Species specific amino acid sequence–protein local structure relationships: An analysis in the light of a structural alphabet

“…Prediction of different features of protein structure like secondary structures (Jones, 1999;Madera et al, 2010;Pollastri et al, 2002), protein disorder (Madera et al, 2010;Xue et al, 2010), transmembrane regions (Illergard et al, 2010;Pylouster et al, 2010), phosphorylation sites (Biswas et al, 2010), protein flexibility (Bornot et al, 2011) or the generation of structural models (de Brevern, 2010;Kelley and Sternberg, 2009), are mainly based on machine learning algorithms (Brylinski and Skolnick, 2008;Rangwala et al, 2009;Xu et al, 2008). Protein structure analyses and prediction methods derive information from non-redundant databanks that represent the state-of-the-art of available data (i.e., solved protein structures).…”