String Kernels and High-Quality Data Set for Improved Prediction of Kinked Helices in α-Helical Membrane Proteins

Abstract

The reasons for distortions from optimal α-helical geometry are widely unknown, but their influences on structural changes of proteins are significant. Hence, their prediction is a crucial problem in structural bioinformatics. For the particular case of kink prediction, we generated a data set of 132 membrane proteins containing 1014 manually labeled helices and examined the environment of kinks. Our sequence analysis confirms the great relevance of proline and reveals disproportionately high occurrences of glycine and serine at kink positions. The structural analysis shows significantly different solvent accessible surface area mean values for kinked and nonkinked helices. More important, we used this data set to validate string kernels for support vector machines as a new kink prediction method. Applying the new predictor, about 80% of all helices could be correctly predicted as kinked or nonkinked even when focusing on small helical fragments. The results exceed recently reported accuracies of alternative approaches and are a consequence of both the method and the data set.

Helix kinks are equally prevalent in soluble and membrane proteins

“…In addition to Proline, studies have suggested other amino acids that could be important in kink occurrence, but none have been consistently identified. For example, Glycine has been found to be prevalent around kinks in some studies, but not in others . Similarly, Serine has been identified as a factor in kinks in some studies, but not in all .…”

“…While some methods have a binary classification of helices (kinked/straight), others have a ternary system (kinked/curved/straight or kinked/distorted/straight). The algorithms used to define the helix set also differ, these studies use methods such as DSSP, the PDB annotation, or manual inspection . As a consequence, the methods do not consistently identify the same set of helices as kinked.…”

“…The number of kinked helices identified differs from study to study; ranging from 6% kinked (although a further 59% of helices are annotated as curved), through 44%, 45% (19% curved), and 60%, to 64% kinked . After kink annotation in transmembrane helices it is then possible to investigate residue types around a kink (e.g., the presence of Proline), or other indicators of kinks (e.g., absence of a backbone hydrogen bond).…”

“…After kink annotation in transmembrane helices it is then possible to investigate residue types around a kink (e.g., the presence of Proline), or other indicators of kinks (e.g., absence of a backbone hydrogen bond). Such characteristics can be used to predict which helices are kinked …”

Statistical analyses and computational prediction of helical kinks in membrane proteins

“…Meruelo et al [24] proposed a neural network approach, TMKink, and achieved a result with sensitivity and specificity of 0.7 and 0.89, respectively. Recently, Kneissl et al [25] suggested the use of string kernels for support vector machines to predict kink positions, which showed about 80 % of all helices can be correctly predicted as kinked or non-kinked. However these methods fail to provide reliable predictions of helical kink angles.…”