出版社：Academy & Industry Research Collaboration Center (AIRCC)
摘要：According to the principle of similar property, structurally similar compounds exhibit very similarproperties and, also, similar biological activities. Many researchers have applied this principle todiscovering novel drugs, which has led to the emergence of the chemical structure-based activityprediction. Using this technology, it becomes easier to predict the activities of unknowncompounds (target) by comparing the unknown target compounds with a group of already knownchemical compounds. Thereafter, the researcher assigns the activities of the similar and knowncompounds to the target compounds. Various Machine Learning (ML) techniques have been usedfor predicting the activity of the compounds. In this study, the researchers have introduced a novelpredictive system, i.e., MaramalNet, which is a convolutional neural network that enables theprediction of molecular bioactivities using a different molecular matrix representation.MaramalNet is a deep learning system which also incorporates the substructure information withregards to the molecule for predicting its activity. The researchers have investigated this novelconvolutional network for determining its accuracy during the prediction of the activities for theunknown compounds. This approach was applied to a popular dataset and the performance of thissystem was compared with three other classical ML algorithms. All experiments indicated thatMaramalNet was able to provide an interesting prediction rate (where the highly diverse datasetshowed 88.01% accuracy, while a low diversity dataset showed 99% accuracy). Also,MaramalNet was seen to be very effective for the homogeneous datasets but showed a lowerperformance in the case of the structurally heterogeneous datasets.
