RF-MaloSite

RF-MaloSite predicts malonylation sites in protein sequences using machine learning and deep learning for accurate site identification.

Key Features:

• RF-MaloSite: Integrates biochemical, physiochemical, and sequence-based features with random forest algorithms.
• DL-MaloSite: Applies deep learning to primary amino acid sequences as input.
• Input data: Uses FASTA-formatted protein sequences represented as 29-mer protein windows.
• RF-MaloSite performance: Achieved MCC scores of 0.42 in cross-validation and 0.40 on an independent test.
• DL-MaloSite performance: Achieved MCC scores of 0.51 in cross-validation and 0.49 on an independent test.
• Computational efficiency: Computational efficiency matches or exceeds existing malonylation prediction tools.

Scientific Applications:

• Malonylation Research: Predicts sites linked to cardiovascular disease, cancer, and crosstalk with acetylation, glutarylation, and succinylation.

Methodology:

Combines random forest with diverse feature sets and deep learning with sequence inputs for malonylation site prediction using FASTA datasets with 29-mer protein windows.