UMPred-FRL: A New Approach for Accurate Prediction of Umami Peptides Using Feature Representation Learning

Abstract

Umami ingredients have been identified as important factors in food seasoning and production. Traditional experimental methods for characterizing peptides exhibiting umami sensory properties (umami peptides) are time-consuming, laborious, and costly. As a result, it is preferable to develop computational tools for the large-scale identification of available sequences in order to identify novel peptides with umami sensory properties. Although a computational tool has been developed for this purpose, its predictive performance is still insufficient. In this study, we use a feature representation learning approach to create a novel machine-learning meta-predictor called UMPred-FRL for improved umami peptide identification. We combined six well-known machine learning algorithms (extremely randomized trees, k-nearest neighbor, logistic regression, partial least squares, random forest, and support vector machine) with seven different feature encodings (amino acid composition, amphiphilic pseudo-amino acid composition, dipeptide composition, composition-transition-distribution, and pseudo-amino acid composition) to develop the final meta-predictor. Extensive experimental results demonstrated that UMPred-FRL was effective and achieved more accurate performance on the benchmark dataset compared to its baseline models, and consistently outperformed the existing method on the independent test dataset. Finally, to aid in the high-throughput identification of umami peptides, the UMPred-FRL web server was established and made freely available online. It is expected that UMPred-FRL will be a powerful tool for the cost-effective large-scale screening of candidate peptides with potential umami sensory properties.

Keywords: bioinformatics; feature representation learning; machine learning; sequence analysis; umami peptide.

Figure 1

The overall flowchart of the development of UMPred-FRL. It consists of dataset construction, feature extraction, baseline model construction, new feature representation generation, and a final meta-predictor development.

Figure 2

Performance comparison of different baseline models. (A,B) Cross-validation and (C,D) independent test results of 42 baseline models. (A,C) The performance of 42 baseline models in terms of cross-validation and independent test ACC. (B,D) The average performance of each classifier over seven different feature descriptors on the training and independent test datasets, respectively.

Figure 3

Performance evaluations of top 30 baseline models. (A,B) Cross-validation BACC and MCC of top 30 baseline models. (C,D) Independent test BACC and MCC of top 30 baseline models.

Figure 4

t-distributed stochastic neighbor embedding (t-SNE) distribution of the positive and negative samples on the training (A–C) and independent test (D–F) datasets, respectively. (A,D) AAC, (B,E) CTDC and (C,F) optimal PF.

Figure 5

SHAP values of informative 7 probabilistic features used for UMPred-FRL. SHAP values represent the directionality of the informative features, where positive and negative SHAP values represent positive (umami peptide) and negative (non-umami peptide) predictions.

Figure 6

Performance comparison of UMPred-FRL with the top five baseline models on the training (A,B) and independent test (C,D) datasets. Prediction results of UMPred-FRL and the top five baseline models in terms of ACC, BACC, Sn, Sp, and MCC. (C,D) ROC curves and AUC values of the top five baseline models.

Figure 7

Performance of the proposed UMPred-FRL and the existing method (iUmami-SCM) on training (A,B) and independent test (C,D) datasets. (A,B) Prediction results of UMPred-FRL and iUmami-SCM in terms of ACC, BACC, Sn, Sp and MCC. (C,D) ROC curves and AUC values of UMPred-FRL and iUmami-SCM.