Length-dependent prediction of protein intrinsic disorder

Abstract

Background: Due to the functional importance of intrinsically disordered proteins or protein regions, prediction of intrinsic protein disorder from amino acid sequence has become an area of active research as witnessed in the 6th experiment on Critical Assessment of Techniques for Protein Structure Prediction (CASP6). Since the initial work by Romero et al. (Identifying disordered regions in proteins from amino acid sequences, IEEE Int. Conf. Neural Netw., 1997), our group has developed several predictors optimized for long disordered regions (>30 residues) with prediction accuracy exceeding 85%. However, these predictors are less successful on short disordered regions (< or =30 residues). A probable cause is a length-dependent amino acid compositions and sequence properties of disordered regions.

Results: We proposed two new predictor models, VSL2-M1 and VSL2-M2, to address this length-dependency problem in prediction of intrinsic protein disorder. These two predictors are similar to the original VSL1 predictor used in the CASP6 experiment. In both models, two specialized predictors were first built and optimized for short (< or = 30 residues) and long disordered regions (>30 residues), respectively. A meta predictor was then trained to integrate the specialized predictors into the final predictor model. As the 10-fold cross-validation results showed, the VSL2 predictors achieved well-balanced prediction accuracies of 81% on both short and long disordered regions. Comparisons over the VSL2 training dataset via 10-fold cross-validation and a blind-test set of unrelated recent PDB chains indicated that VSL2 predictors were significantly more accurate than several existing predictors of intrinsic protein disorder.

Conclusion: The VSL2 predictors are applicable to disordered regions of any length and can accurately identify the short disordered regions that are often misclassified by our previous disorder predictors. The success of the VSL2 predictors further confirmed the previously observed differences in amino acid compositions and sequence properties between short and long disordered regions, and justified our approaches for modelling short and long disordered regions separately. The VSL2 predictors are freely accessible for non-commercial use at http://www.ist.temple.edu/disprot/predictorVSL2.php.

Figure 1

VSL2 predictor architectures. The final prediction for VSL2-M1 is calculated as O_L × O_M + O_S × (1 – O_M), while for VSL2-M2 it is the output of meta predictor M2. The inputs for M2 are 2 × W_in predictions by VSL2-L and VSL2-S for the neighbouring residues in a window of length W_in. All component predictors are built using classification algorithms that approximate the posterior probability p(c = 1|x), where x is the feature (input) vector and c is the class label.

Figure 2

Comparison of amino acid compositions between short and long disordered regions. The y-axis represents the difference in amino acid compositions (fractions) from a reference dataset of ordered proteins, Globular-3D. The error bars correspond to one standard deviation estimated using 5,000 bootstrap samples. His-tags and initial methionines were not counted.

Figure 3

Length-dependent prediction accuracies. Per-residue accuracies (sensitivities) are reported on disordered regions from different length ranges.

Figure 4

Representative predictions on two PDB chains. (A) 1REP:C with four short disordered regions at residue 1–14, 50–55, 98–109, and 247–251. (B) 1B70:A with a long disordered region at residue 1–85. These disordered regions are marked as thick line segments. Residues with predictions above 0.5 are interpreted as predicted disordered.

Figure 5

Comparison of receiver operating characteristic (ROC) curves. The ROC curves were plotted using (A) per-chain and (B) per-residue accuracies, by varying the decision thresholds from 0 to 1 in increments of 0.001. The corresponding AUC values were approximated using the trapezoid rule and reported in Table 7.

Figure 6

VSL2 prediction on PDB chain 1 YYH:B. VSL2 prediction (disorder probability) is plotted in blue sold line. Residues with predictions above 0.5 are interpreted as predicted disordered. The long region of missing electron density (residues 1–54) is marked as thick red segment. The fourteen short green segments correspond to the α-helices in the seven ANK repeats (two helices for each repeat).