Genome Sequence Analysis and Drug-Target Interaction Prediction Using Deep Learning

In modern biology, research in the fields of genome sequences and drug-target interaction predictions is booming. A deeper understanding of genetic information has been made possible by the analyzing and decoding of the human genomes and the development of advanced sequencing methods. Predicting drug-target interactions is a critical aspect in drug development. Experimenting and developing new ways to discover such relationships based on treatments are costly, complex, and time-consuming. When we consider the whole time and money involved, inventing new, more accurate computational methods may be more efficient than experimental ones. This research is based on a deep learning approach to analyze the genome sequences and accurately predict how the drugs would interact with the target sequences. It focuses on using long short-term memory (LSTM) networks that makes use of the -mer encoding technique for sequence analysis and predicting drug-target interactions. Analyzing genome sequences help in identifying patterns and commonalities of the human DNA. For more accurate predictions, the drug-target interaction prediction model developed uses DeepLSTM combined with principal component analysis (PCA) for feature reduction and Position Specific Scoring Matrix (PSSM) for feature extraction. On the basis of our work, the proposed methodology can be used to understand and detect novel drug-target interactions as well as genome clusters. This research contributes to understanding of human DNA and also provides a robust framework for drug-target identification.