SPA: Short peptide analyzer of intrinsic disorder status of short peptides

Abstract

Disorder prediction for short peptides is important and difficult. All modern predictors have to be optimized on a preselected dataset prior to prediction. In the succeeding prediction process, the predictor works on a query sequence or its short segment. For implementing the prediction smoothly and obtaining sound prediction results, a specific length of the sequence or segment is usually required. The need of the preselected dataset in the optimization process and the length limitation in the prediction process restrict predictors' performance. To minimize the influence of these limitations, we developed a method for the prediction of intrinsic disorder in short peptides based on large dataset sampling and statistics. As evident from the data analysis, this method provides more reliable prediction of the intrinsic disorder status of short peptides.

Figure 1

Length distribution of short disordered and short ordered segments from DSP and OSP.

Figure 2

Relative composition profile of DSP segments vs FDD (a) and that of OSP segments vs fully ordered dataset (FOD) (b). On x-axis, amino acids are arranged in ascending disorder tendency. C_P is the absolute composition of one amino acid in the query dataset; C_ID is the absolute composition of the same amino acid in FDD; C_OD is the same amino acid composition in FOD. Error bars are from 200 times of bootstrapping sampling. ‘D’ indicates subsets from DSP while ‘O’ is for subsets from OSP. D5 includes all the segments with segment length less than 5; D10 is for segments longer than or equal to 5 but <10; D15 corresponds to segments with 10 ≤ L < 15; D20 is 15 ≤ L < 20; D25 is 20 ≤ L < 25; D28 is 25 ≤ L ≤ 28. The same nomenclature is applied to subsets obtained from OSP.

Figure 3

Receiver operating characteristic (ROC) curve of SPA in DSP/OSP datasets. These two datasets are furthermore grouped into subsets according to the length of segments in them. ‘D’ indicates subsets from DSP while ‘O’ is for subsets from OSP. D5 includes all the segments with segment length less than 5; D10 is for segments longer than or equal to five but <10; D15 corresponds to segments with 10 ≤ L < 15; D20 is 15 ≤ L < 20; D25 is 20 ≤ L < 25; D28 is 25 ≤ L ≤ 28. The same nomenclature is applied to subsets obtained from OSP. Each pair of subsets with the same range of length, originated from DSP and OSP, respectively, are put together to calculate the ROC curve for segments of that length.

Figure 4

3D Structure of molecular recognition features (MoRFs) with their substrates. (a) (PDBid:2NM1) and (b) (PDBid:2AUC) are alpha-MoRFs. (a) is predicted to be structured while (b) is disordered. (c) (PDBid:1LXH) and (d) (PDNid: 1PJM) are coil-MoRFs with (c) structured and (d) disordered.

Figure 5

Application of SPA on two peptides P1 (PFVVSDIAFMGLFYD) and P2 (PLSHGSVVYPRSSLG). P1 is experimentally identified as ordered while P2 is disordered. All the slim curves are PONDR-VLXT predictions for the combined peptides by inserting the query peptides into disordered and ordered segments selected from fully disordered segments (FDS) and fully ordered segments (FOS). The large connected dots are predictions and error bars from SPA. (a) Predictions of 10 randomly selected combined peptides by embedding P1 on disordered protein segments from FDS. (b) Predictions of 10 randomly combined peptides by implanting P1 into ordered segments of FOS. (c) Predictions for peptides generated by inserting P2 into 10 segments used in (a). (d) Combining P2 onto 10 segments shown in (b).