Integration of A Deep Learning Classifier with A Random Forest Approach for Predicting Malonylation Sites

Area of Science:

• Biochemistry
• Bioinformatics
• Computational Biology

Background:

• Protein malonylation is an emerging post-translational modification impacting diverse biological processes.
• Accurate identification of malonylation sites is essential for mechanistic studies.
• Existing prediction methods have limitations in accuracy and false positive rates.

Purpose of the Study:

• To develop a highly accurate computational tool for predicting mammalian protein malonylation sites.
• To improve upon existing prediction methods for malonylation site identification.
• To provide a reliable tool for researchers investigating protein malonylation.

Main Methods:

• Construction of a deep learning (DL) classifier using long short-term memory with word embedding (LSTMWE).
• Integration of LSTMWE with a random forest classifier employing enhanced amino acid content encoding, termed LEMP.
• Comparison of LEMP against traditional classifiers and existing malonylation predictors.

Main Results:

• LSTMWE demonstrated superior performance compared to traditional feature encoding methods and DL classifiers using one-hot vectors.
• The integrated LEMP approach outperformed individual classifiers and current state-of-the-art malonylation predictors.
• LEMP achieved a promisingly low false positive rate, enhancing its predictive utility.

Conclusions:

• LEMP is an effective and reliable tool for identifying protein malonylation sites with high confidence.
• The integration of DL and ML approaches offers a powerful strategy for biological site prediction.
• LEMP is publicly available, facilitating research in protein malonylation.