Prediction of protein N-formylation and comparison with N-acetylation based on a feature selection method

Formator: Predicting Lysine Formylation Sites Based on the Most Distant Undersampling and Safe-Level Synthetic Minority Oversampling

Lysine formylation is a reversible type of protein post-translational modification and has been found to be involved in a myriad of biological processes, including modulation of chromatin conformation and gene expression in histones and other nuclear proteins. Accurate identification of lysine formylation sites is essential for elucidating the underlying molecular mechanisms of formylation. Traditional experimental methods are time-consuming and expensive. As such, it is desirable and necessary to develop computational methods for accurate prediction of formylation sites. In this study, we propose a novel predictor, termed Formator, for identifying lysine formylation sites from sequences information. Formator is developed using the ensemble learning (EL) strategy based on four individual support vector machine classifiers via a voting system. Moreover, the most distant undersampling and Safe-Level-SMOTE oversampling techniques were integrated to deal with the data imbalance problem of the training dataset. Four effective feature extraction methods, namely bi-profile Bayes (BPB), k-nearest neighbor (KNN), amino acid physicochemical properties (AAindex), and composition and transition (CTD) were employed to encode the surrounding sequence features of potential formylation sites. Extensive empirical studies show that Formator achieved the accuracy of 87.24 and 74.96 percent on jackknife test and the independent test, respectively. Performance comparison results on the independent test indicate that Formator outperforms current existing prediction tool, LFPred, suggesting that it has a great potential to serve as a useful tool in identifying novel lysine formylation sites and facilitating hypothesis-driven experimental efforts.

A Transfer Learning-Based Approach for Lysine Propionylation Prediction

Lysine propionylation is a newly discovered posttranslational modification (PTM) and plays a key role in the cellular process. Although proteomics techniques was capable of detecting propionylation, large-scale detection was still challenging. To bridge this gap, we presented a transfer learning-based method for computationally predicting propionylation sites. The recurrent neural network-based deep learning model was trained firstly by the malonylation and then fine-tuned by the propionylation. The trained model served as feature extractor where protein sequences as input were translated into numerical vectors. The support vector machine was used as the final classifier. The proposed method reached a matthews correlation coefficient (MCC) of 0.6615 on the 10-fold crossvalidation and 0.3174 on the independent test, outperforming state-of-the-art methods. The enrichment analysis indicated that the propionylation was associated with these GO terms (GO:0016620, GO:0051287, GO:0003735, GO:0006096, and GO:0005737) and with metabolism. We developed a user-friendly online tool for predicting propoinylation sites which is available at http://47.113.117.61/.

The Blood Gene Expression Signature for Kawasaki Disease in Children Identified with Advanced Feature Selection Methods

Kawasaki disease (KD) is an acute vasculitis, accompanied by coronary artery aneurysm, coronary artery dilatation, arrhythmia, and other serious cardiovascular diseases. So far, the etiology of KD is unclear; it is necessary to study the molecular mechanism and related factors of KD. In this study, we analyzed the expression profiles of 75 DB (identifying bacteria), 122 DV (identifying virus), 71 HC (healthy control), and 311 KD (Kawasaki disease) samples. 332 key genes related to KD and pathogen infections were identified using a combination of advanced feature selection methods: (1) Boruta, (2) Monte-Carlo Feature Selection (MCFS), and (3) Incremental Feature Selection (IFS). The number of signature genes was narrowed down step by step. Subsequently, their functions were revealed by KEGG and GO enrichment analyses. Our results provided clues of potential molecular mechanisms of KD and were helpful for KD detection and treatment.

Identification of Gene Signatures and Expression Patterns During Epithelial-to-Mesenchymal Transition From Single-Cell Expression Atlas

Cancer, which refers to abnormal cell proliferative diseases with systematic pathogenic potential, is one of the leading threats to human health. The final causes for patients’ deaths are usually cancer recurrence, metastasis, and drug resistance against continuing therapy. Epithelial-to-mesenchymal transition (EMT), which is the transformation of tumor cells (TCs), is a prerequisite for pathogenic cancer recurrence, metastasis, and drug resistance. Conventional biomarkers can only define and recognize large tissues with obvious EMT markers but cannot accurately monitor detailed EMT processes. In this study, a systematic workflow was established integrating effective feature selection, multiple machine learning models [Random forest (RF), Support vector machine (SVM)], rule learning, and functional enrichment analyses to find new biomarkers and their functional implications for distinguishing single-cell isolated TCs with unique epithelial or mesenchymal markers using public single-cell expression profiling. Our discovered signatures may provide an effective and precise transcriptomic reference to monitor EMT progression at the single-cell level and contribute to the exploration of detailed tumorigenesis mechanisms during EMT.

The Serum MicroRNA Signatures for Pancreatic Cancer Detection and Operability Evaluation

Pancreatic cancer (PC) has high morbidity and mortality. It is the fourth leading cause of cancer death. Its diagnosis and treatment are difficult. Liquid biopsy makes early diagnosis of pancreatic cancer possible. We analyzed the expression profiles of 2,555 serum miRNAs in 100 pancreatic cancer patients and 150 healthy controls. With advanced feature selection methods, we identified 13 pancreatic cancer signature miRNAs that can classify the pancreatic cancer patients and healthy controls. For pancreatic cancer treatment, operation is still the first choice. But many pancreatic cancer patients are already inoperable. Therefore, we compared the 79 inoperable and 21 operable patients and identified 432 miRNAs that can predict whether a pancreatic cancer patient was operable. The functional analysis of the 13 pancreatic cancer signatures and the 432 operability miRNAs revealed the molecular mechanisms of pancreatic cancer and shield light on the diagnosis and therapy of pancreatic cancer in clinical practice.

An Information Entropy-Based Approach for Computationally Identifying Histone Lysine Butyrylation

Butyrylation plays a crucial role in the cellular processes. Due to limit of techniques, it is a challenging task to identify histone butyrylation sites on a large scale. To fill the gap, we propose an approach based on information entropy and machine learning for computationally identifying histone butyrylation sites. The proposed method achieves 0.92 of area under the receiver operating characteristic (ROC) curve over the training set by 3-fold cross validation and 0.80 over the testing set by independent test. Feature analysis implies that amino acid residues in the down/upstream of butyrylation sites would exhibit specific sequence motif to a certain extent. Functional analysis suggests that histone butyrylation was most possibly associated with four pathways (systemic lupus erythematosus, alcoholism, viral carcinogenesis and transcriptional misregulation in cancer), was involved in binding with other molecules, processes of biosynthesis, assembly, arrangement or disassembly and was located in such complex as consists of DNA, RNA, protein, etc. The proposed method is useful to predict histone butyrylation sites. Analysis of feature and function improves understanding of histone butyrylation and increases knowledge of functions of butyrylated histones.