Disulfide Connectivity Prediction Based on Modelled Protein 3D Structural Information and Random Forest Regression

Abstract

Disulfide connectivity is an important protein structural characteristic. Accurately predicting disulfide connectivity solely from protein sequence helps to improve the intrinsic understanding of protein structure and function, especially in the post-genome era where large volume of sequenced proteins without being functional annotated is quickly accumulated. In this study, a new feature extracted from the predicted protein 3D structural information is proposed and integrated with traditional features to form discriminative features. Based on the extracted features, a random forest regression model is performed to predict protein disulfide connectivity. We compare the proposed method with popular existing predictors by performing both cross-validation and independent validation tests on benchmark datasets. The experimental results demonstrate the superiority of the proposed method over existing predictors. We believe the superiority of the proposed method benefits from both the good discriminative capability of the newly developed features and the powerful modelling capability of the random forest. The web server implementation, called TargetDisulfide, and the benchmark datasets are freely available at: http://csbio.njust.edu.cn/bioinf/TargetDisulfide for academic use.

Residue Adjacency Matrix Based Feature Engineering for Predicting Cysteine Reactivity in Proteins

“…A comparison between the different values of n was performed and results reported in Table 8 . The n of 6 was found to be effective in prior research [ 36 ] It is a coincidence that RAMseq and RAMmod were fairly close to the number 6 for PSS and PSSM as seen in Table 2 . We used 6 as a general number that can be expected to perform well on any new dataset based on the results.…”

“…For a window size of 2*k + 1 (k positions to the left of the target cysteine, k to the right, and the target cysteine position itself) and the twenty major amino acids, we get a matrix that is twenty rows long and 2*k + 1 columns wide that can be vectorized for a total of 20*(2*k + 1) features. We chose 6 for k as found in the prior work [ 36 ]. Our PSSMs provided 260 entries for the classifier.…”

“…We use a window size of thirteen positions (the target cysteine plus the six positions to the left and the six positions to the right) for the PSS matrix. 6 was chosen due to prior works [ 36 ]. The final matrix is then thirteen by three (the confidence score for the three classes of secondary structures) which results in a thirty-nine-dimensional feature source for the classifier once vectorized.…”

A New Semisupervised-Entropy Framework of Hyperspectral Image Classification Based on Random Forest

“…The classification results are voted on by the decision trees. The final feature category relies on the classification result with the largest number of votes [32]. The Gini index is used to measure the classification results.…”

Accurate disulfide-bonding network predictions improveab initiostructure prediction of cysteine-rich proteins

“…Different from the above ab inito approaches which perform predictions by only using the amino acid sequence information, the other trend is using the homology modeling techniques, where some prediction features are extracted from the modeled structures. For instance, the spatial distance between the cysteine residues in the modeled structure can be used as an encoding feature (Yu et al, 2015). Other studies in this trend include: Lin and Tseng (2010) and O' Connor and Yeates (2004).…”