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

Abstract

As a newly-identified protein post-translational modification, malonylation is involved in a variety of biological functions. Recognizing malonylation sites in substrates represents an initial but crucial step in elucidating the molecular mechanisms underlying protein malonylation. In this study, we constructed a deep learning (DL) network classifier based on long short-term memory (LSTM) with word embedding (LSTMWE) for the prediction of mammalian malonylation sites. LSTMWE performs better than traditional classifiers developed with common pre-defined feature encodings or a DL classifier based on LSTM with a one-hot vector. The performance of LSTMWE is sensitive to the size of the training set, but this limitation can be overcome by integration with a traditional machine learning (ML) classifier. Accordingly, an integrated approach called LEMP was developed, which includes LSTMWE and the random forest classifier with a novel encoding of enhanced amino acid content. LEMP performs not only better than the individual classifiers but also superior to the currently-available malonylation predictors. Additionally, it demonstrates a promising performance with a low false positive rate, which is highly useful in the prediction application. Overall, LEMP is a useful tool for easily identifying malonylation sites with high confidence. LEMP is available at http://www.bioinfogo.org/lemp.

DeepCSO: a deep-learning network approach to predicting Cysteine S-sulphenylation sites

“…The EAAC encoding [26,[32][33][34] introduces a fixed-length sliding window based on the encoding of Amino Acid Composition (AAC), which calculates the frequency of each type of amino acids in a protein or peptide sequence [35]. EAAC is calculated by continuously sliding a fixed-length sequence window (using the default value as 5) from the N-terminus to the C-terminus of each peptide.…”

DeepKhib: a deep-learning framework for lysine 2-hydroxyisobutyrylation sites prediction

“…Khib sites were considered positives whereas the remaining K sites were taken as negatives. We further estimated the potential redundancy of the positive sites by extracting the peptide segment of seven residues with the Khib site in the center and count the number of unique segments [20,25]. The number (9,444) through ten-fold cross-validation ( Fig.…”

“…The DL models included a Gated Recurrent Unit (GRU) model with the word-embedding encoding approach dubbed GRUWE and two CNN models with the one-hot and word-embedding encoding approaches named CNNOH and CNNWE, respectively. Both encoding methods are common in the DL algorithms [20,25].…”

“…Therefore, the general model may be effectually applied to any species. Furthermore, we evaluated the generality of the general CNNOH model using the dataset of S. cerevisiae that contained 1,049 positive and 1,049 negative samples, which may not be enough for build an effective DL predictor [20]. The general model got the AUC value as 0.789, indicating the generality of this model.…”

“…Recently, the deep learning (DL) algorithms, as the modern ML architecture, have demonstrated superior prediction performance in the field of bioinformatics, such as the prediction of modification sites on DNA, RNA and proteins [15][16][17][18][19]. We have developed a few DL approaches for the prediction of PTM sites and they all demonstrate their superiority over conventional ML algorithms [20][21][22]. Therefore, we attempted to compare the DL models with the traditional ML models for the prediction of Khib sites.…”