The assessment of efficient representation of drug features using deep learning for drug repositioning

Abstract

BackgroundDe novo drug discovery is a time-consuming and expensive process. Nowadays, drug repositioning is utilized as a common strategy to discover a new drug indication for existing drugs. This strategy is mostly used in cases with a limited number of candidate pairs of drugs and diseases. In other words, they are not scalable to a large number of drugs and diseases. Most of the in-silico methods mainly focus on linear approaches while non-linear models are still scarce for new indication predictions. Therefore, applying non-linear computational approaches can offer an opportunity to predict possible drug repositioning candidates.ResultsIn this study, we present a non-linear method for drug repositioning. We extract four drug features and two disease features to find the semantic relations between drugs and diseases. We utilize deep learning to extract an efficient representation for each feature. These representations reduce the dimension and heterogeneity of biological data. Then, we assess the performance of different combinations of drug features to introduce a pipeline for drug repositioning. In the available database, there are different numbers of known drug-disease associations corresponding to each combination of drug features. Our assessment shows that as the numbers of drug features increase, the numbers of available drugs decrease. Thus, the proposed method with large numbers of drug features is as accurate as small numbers.ConclusionOur pipeline predicts new indications for existing drugs systematically, in a more cost-effective way and shorter timeline. We assess the pipeline to discover the potential drug-disease associations based on cross-validation experiments and some clinical trial studies.

Evaluating Performance of Drug Repurposing Technologies

“…The drug repurposing project PREDICT uses AUROC as a main method of reporting goodness [60]. Moridi et al, report their own AUROC compared to that of PREDICT [61]. Nguyen et al create a computational drug repurposing framework based on control system theory (DeCoST) to make novel treatment predictions for cancer [62].…”

Successful stories of drug repurposing for cancer therapy in hepatocellular carcinoma

“…Drug discovery process involves multidisciplinary integrating experiences starting with in silico computational modeling to design scaffolds of interest based on the affiliated therapeutic molecular targets to different clinical diseases. Typically, this is followed by synthetic methodologies optimization to assemble final ligands ready for biological evaluation [ 72 ]. Preclinical biological evaluations involving in vitro and in vivo assessments create significant portion for getting the potential ligands to be considered as potential drug candidates or active pharmaceutical ingredients; however, there are different discovery stories that failed to make it into the market due to intellectual property, efficacy, safety, toxicity, biopharmaceutical compatibility, cost/benefit feasibility, regulatory affairs, etc.…”

“…This strategy helped potential drugs, such as metformin, digoxin, and statins to be repurposed for different types of cancer. The integration of computational tools has the adequate share for the advancement and validation of the drug repositioning strategy to evolve in silico computational screening–based repositioning [ 72 ]. This strategy helped the repositioning of nonsteroidal antiinflammatory drugs and proton-pump inhibitors for different types of cancer [ 78 , 79 ].…”

A Drug Repositioning Algorithm Based on Deep Auto-Encoder and Adaptive Fusion

“…And "Adagrad" [14] is regarded as an optimization method. The encoder consists of an input layer and three building layers [15] . To be specific, the building layer is composed of a fully connected layer and a discarded layer, and the last building layer is a coding layer.…”