T4SE-XGB: interpretable sequence-based prediction of type IV secreted effectors using eXtreme gradient boosting algorithm

Abstract

Type IV secreted effectors (T4SEs) can be translocated into the cytosol of host cells via type IV secretion system (T4SS) and cause diseases. However, experimental approaches to identify T4SEs are time-and resource-consuming, and the existing computational tools based on machine learning techniques have some obvious limitations such as the lack of interpretability in the prediction models. In this study, we proposed a new model, T4SE-XGB, which uses the eXtreme gradient boosting (XGBoost) algorithm for accurate identification of type IV effectors based on optimal protein sequence features. After trying 20 different features, the best result achieved when all features were fed into XGBoost by the 5-fold cross validation compared with different machine learning methods. Then, the ReliefF algorithm was adopted to optimize feature vectors and got final 1100 features for our dataset which obviously improved the model performance. T4SE-XGB exhibited highest predictive performance on the independent test set and clearly outperforms other recent prediction tools. What's more, the SHAP method was used to interpret the contribution of features to model predictions. The identification of key features can contribute to an improved understanding of multifactorial contributors to host-pathogen interactions and bacterial pathogenesis. In addition to type IV effector prediction, we believe that the proposed framework composed of model construction and model interpretation can provide more instructive guidance for further research of developing novel computational methods and mechanism exploration of biological problems. The data and code for this study can be found at https://github.com/CT001002/T4SE-XGB.

Tracking Major Sources of Water Contamination Using Machine Learning

“…In traditional gradient boosting, each new tree specifically focuses on the error of the previous tree. XGBoost adds more regularization terms in the model to control model over-fitting, which makes the model have a better performance ( Chen and Guestrin, 2016 ; Chen et al, 2020 ). In this study, “XGBClassifier” from “xgboost” library 7 was used for prediction.…”

Computational prediction of secreted proteins in gram-negative bacteria

“…CNN-T4SE integrated three Convolutional Neural Network models training the amino acid composition, solvent accessibility and secondary structure of full-length T4SEs, achieving better performance than other tools and lower false positive predictions [298] . Other groups adopted an alternative strategy, by selecting the best optimized features, and/or training and identifying the best machine learning models, to improve the prediction performance [299] , [149] , [150] , [151] . Some of the models have been well applied in identification of T4SEs in L. pneumophila [151] and Anaplasma phagocytophilum (OPT4e; [150] ).…”

FPredX: Interpretable models for the prediction of spectral maxima, brightness, and oligomeric states of fluorescent proteins

“…Importance measure of every aligned position to the predictive performance on test sets of each model was calculated by the SHAP package, 41 which was frequently adopted to understand sequence-property relationship in proteins. [42][43][44] For every one-hot feature of an FP sequence that undergoes prediction, SHAP assigns an importance measure to the feature called the SHAP value. A positive SHAP value corresponds to a positive contribution of the feature value to the predicted target, while a higher SHAP value corresponds to a higher importance of the feature value to the prediction of the target.…”