Deep Learning for Drug Discovery: A Study of Identifying High Efficacy Drug Compounds Using a Cascade Transfer Learning Approach

In this research, we applied deep learning to rank the effectiveness of candidate drug compounds in combating viral cells, in particular, SARS-Cov-2 viral cells. For this purpose, two different datasets from Recursion Pharmaceuticals, a siRNA image dataset (RxRx1), which were used to build and calibrate our model for feature extraction, and a SARS-CoV-2 dataset (RxRx19a) was used to train our model for ranking efficacy of candidate drug compounds. The SARS-CoV-2 dataset contained healthy, uninfected control or “mock” cells, as well as “active viral” cells (cells infected with COVID-19), which were the two cell types used to train our deep learning model. In addition, it contains viral cells treated with different drug compounds, which were the cells not used to train but test our model. We devised a new cascade transfer learning strategy to construct our model. We first trained a deep learning model, the DenseNet, with the siRNA set, a dataset with characteristics similar to the SARS-CoV-2 dataset, for feature extraction. We then added additional layers, including a SoftMax layer as an output layer, and retrained the model with active viral cells and mock cells from the SARS-CoV-2 dataset. In the test phase, the SoftMax layer outputs probability (equivalently, efficacy) scores which allows us to rank candidate compounds, and to study the performance of each candidate compound statistically. With this approach, we identified several compounds with high efficacy scores which are promising for the therapeutic treatment of COVID-19. The compounds showing the most promise were GS-441524 and then Remdesivir, which overlapped with these reported in the literature and with these drugs that are approved by FDA, or going through clinical trials and preclinical trials. This study shows the potential of deep learning in its ability to identify promising compounds to aid rapid responses to future pandemic outbreaks.

Comprehensive analysis of tumor microenvironment landscape and potential therapeutic agents based on tumor-infiltrating immune cells in colorectal cancer

Background: Tumor-infiltrating immune cells (TIIC) are the major components of the tumor microenvironment (TME) and play vital roles in the tumorigenesis and progression of colorectal cancer (CRC). Increasing evidence has elucidated their significances in predicting prognosis and therapeutic efficacy. Nonetheless, the immune infiltrative landscape of CRC remains largely unknown. Methods: All the RNA-seq transcriptome data and full clinical annotation of 1213 colorectal cancer patients were downloaded from The Cancer Genome Atlas (TCGA) and Gene-Expression Omnibus (GEO) database. The “CIBERSORT” and “estimate” R package were applied to calculate 22 infiltrated immune cell fractions and stromal and immune score. Three TIIC patterns were determined by Unsupervised clustering methods. Through using principal-component analysis, TIIC scores were established. Data for potential agents comes from the Profiling Relative Inhibition Simultaneously in Mixtures (PRISM) and Cancer Therapeutics Response Portal database (CTRP). Results：In this study, we identified three distinct TIIC patterns characterized by distinct immunological features in 1213 CRC samples from multiple platforms. Base on the TIIC-related gene signatures from three clusters, we constructed a scoring system to quantify the immune infiltration level of individual samples in the CRC cohort and the clinical benefits of different groups. The high TIIC score group was marked by increased immune activation status and favorable prognosis. Conversely, low TIIC score group was featured with immune-desert phenotype and poor prognosis, along with the activation of transforming growth factor-β (TGF-β), WNT, ECM receptor interaction, and VEGF signaling pathways. Meanwhile, the high TIIC score group was also correlated with enhanced efficacy of immunotherapy. Additional, four chemotherapy drugs, seven CTRP-derived drug compounds and six PRISM-derived drug compounds were identified as potential drug for CRC among high and low TIIC subgroups.Conclusions: Collectively, as an effective prognostic biomarker and predictive indicator, the TIIC score plays an important role in the evaluation of CRC prognosis and the response of immunotherapy. Investigation of the TIIC patterns might provide us a promising target for improving immunotherapeutic efficacy in CRC.

Computational Chemistry to Repurposing Drugs for the Control of COVID-19

Thus far, in 2021, 219 countries with over 175 million people have been infected by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). SARS-CoV-2 is a positive sense, single-stranded RNA virus, and is the causal agent for coronavirus disease (COVID-19). Due to the urgency of the situation, virtual screening as a computational modeling method offers a fast and effective modality of identifying drugs that may be effective against SARS-CoV-2. There has been an overwhelming abundance of molecular docking against SARS-CoV-2 in the last year. Due to the massive volume of computational studies, this systematic review has been created to evaluate and summarize the findings of existing studies. Herein, we report on computational articles of drugs which target, (1) viral protease, (2) Spike protein-ACE 2 interaction, (3) RNA-dependent RNA polymerase, and (4) other proteins and nonstructural proteins of SARS-CoV-2. Based on the studies presented, there are 55 identified natural or drug compounds with potential anti-viral activity. The next step is to show anti-viral activity in vitro and translation to determine effectiveness into human clinical trials.

Advanced Materials for Immunosensing of Pharmaceutical and Drug Compounds

Real-time and accurate levels of pharmaceuticals undertake critical effects in the therapy process. Thus, reliable detection of pharmaceuticals is important for regulating the proper concentration of them to enhance the effectiveness and to decrease possible side effects. However, the development of new reliable sensory systems is the main prerequisite for the mentioned aims. Immunosensors can be regarded as an effective tool due to their sensitivity and unique specificity originating from the intrinsic nature of the antigen-antibody interaction. This review reports material tendencies in the development of immunosensors for pharmaceuticals (veterinary and human) which have been reported in the last few years. Carbon-based (graphene, graphene oxide, carbon nanotubes, etc.), gold, and magnetic materials are the main materials for the fabrication of pharmaceutical immunosensors. Also, this review reports benefits and limitations on the reported immunosensor and mechanism and analytical performance of the immunoplatforms to address future researches.

Lipid Based Drug Delivery System: A Review

Associated with the old/traditional of method of drug delivery are several limitation ranging from first-pass effect, low tolerance, minimal bioavailability, fluctuation of plasma drug concentration which result to less or no desired effect produced. This call for the demand for a more efficient drug administration technique. Lipid systems are biocompatible, inert and biodegradable, stable and deliver at the target with the desired effect. This paper attempt to describe several types of lipid particles used to deliver drug compounds and their applications as therapeutic agent in treating different clinical condition.

High-dose drug heat map based on organoid array chip for drug selection with high safety and efficacy

An organoid array chip was developed by adopting a micropillar and microwell structure to test safety and efficacy of drugs using high dose drug heat map. In the chip, we encapsulated patient-derived cells in alginate and grow them to maturity for more than 7 days to form cancer organoids. When screening drug compounds in a high-density organoid array due to lack of number of patient-derived cells, changing media without damage of organoids is a very tedious and difficult process. Organoids grown in conventional well plates needed too many cells and were also easily damaged due to multiple pipetting during maintenance culture or during experimental procedures. To solve those problem, we applied a micropillar and microwell structure to the organoid array. We used patient-derived cells from patients with Glioblastoma multiforme (GBM), the most common and lethal form of central nervous system cancer, to validate the array chip performance. After forming more than 100µm-diameter organoids in 12 [[EQUATION]] 36 pillar array chip (25mm [[EQUATION]] 75mm), we tested 70 drug compounds (6 replicates) with high high-dose to find out high safety and efficacy drug candidates. Comparing the drug response of organoids derived from normal cells and cancer cells, we identified four compounds (Dacomitinib, Cediranib, Ly2835219, BGJ398) as drug candidates without toxicity to GBM cells.