N-terminal acetylation is a specific protein modification that occurs only at the N-terminus but plays a significant role in protein stability, folding, subcellular localization and protein-protein interactions. Computational methods enable finding N-terminal acetylated sites from large-scale proteins efficiently. However, limited by the number of the labeled proteins, existing tools only focus on certain subtypes of N-terminal acetylated sites on frequently detected amino acids. For example, NetAcet focuses on alanine, glycine, serine and threonine only, and N-Ace predicts on alanine, glycine, methionine, serine and threonine. With the growth of experimental N-terminal acetylated site data, it is observed that N-terminal protein acetylation occurs on nearly ten types of amino acids. To facilitate comprehensive analysis, we have developed MTNA (Multiple Types of N-terminal Acetylation), a deep learning network capable of accurately predicting N-terminal protein acetylation sites for various amino acids at the N-terminus. MTNA not only outperforms existing tools but also has the capability to identify rare types of N-terminal protein acetylated sites occurring on less studied amino acids.

With the increasing need for public health and drug development, combination therapy has become widely used in clinical settings. However, the risk of unanticipated adverse effects and unknown toxicity caused by drug-drug interactions (DDIs) is a serious public health issue for polypharmacy safety. Traditional experimental methods for detecting DDIs are expensive and time-consuming. Therefore, many computational methods have been developed in recent years to predict DDIs with the growing availability of data and advancements in artificial intelligence. In silico methods have proven to be effective in predicting DDIs, but detecting potential interactions, especially for newly discovered drugs without an existing DDI network, remains a challenge. In this study, we propose a predicting method of DDIs named HAG-DDI based on graph attention networks. We consider the differences in mechanisms between DDIs and add learning of semantic-level attention, which can focus on advanced representations of DDIs. By treating interactions as nodes and the presence of the same drug as edges, and constructing small subnetworks during training, we effectively mitigate potential bias issues arising from limited data availability. Our experimental results show that our method achieves an F1-score of 0.952, proving that our model is a viable alternative for DDIs prediction. The codes are available at: https://github.com/xtnenu/DDIFramework.

The goal of precision oncology is to select more effective treatments or beneficial drugs for patients. The transcription of ‘‘hidden responders’’ which precision oncology often fails to identify for patients is important for revealing responsive molecular states. Recently, a RAS pathway activation detection method based on machine learning and a nature-inspired deep RAS activation pan-cancer has been proposed. However, we note that the activating gene variations found in KRAS, HRAS and NRAS vary substantially across cancers. Besides, the ability of a machine learning classifier to detect which KRAS, HRAS and NRAS gain of function mutations or copy number alterations causes the RAS pathway activation is not clear. Here, we proposed a deep neural network framework for deciphering and identifying pan-cancer RAS pathway activation (DIPRAS). DIPRAS brings a new insight into deciphering and identifying the pan-cancer RAS pathway activation from a deeper perspective. In addition, we further revealed the identification and characterization of RAS aberrant pathway activity through gene ontological enrichment and pathological analysis. The source code is available by the URL https://github.com/zhaoyw456/DIPRAS.