Abstract

In this work, we describe an end-to-end deep learning architecture for Heart Murmur Detection from Phonocardiogram(PCG) recordings as part of The George B. Moody PhysioNet Challenge 2022. Our team, “Team IIITH” received a weighted accuracy score of 0.708 (ranked 19th out of 40 teams) and Challenge cost of 13264 (ranked 22nd out of 39 teams) on the official hidden test set. In our approach, the PCG recordings are first downsampled to 1000 Hz before being passed through a Butterworth’s low and high pass filter to remove baseline wanders and high-frequency noise present in the recordings. The PCG recordings are then broken down into 10-second segments and normalized to bring all trainable samples to the same size. To extract embeddings more efficiently, we built a custom 1-dimensional Residual Network (ResNet) where the 10-second inputs are passed through variable-sized kernel ResNets in parallel, before being concatenated and passed through the next ResNet layer to account for different length dependencies across the PCG signal. The output of this custom ResNet is then fed to a 2-layer feed-forward network for final classification. Cross-Entropy Loss with class weights was employed to account for class imbalance. Our approach obtained a 5-fold Cross-Validation weighted accuracy score of 0.71 and challenge cost score of 12067 on the training set.

Abstract

Nucleocytoplasmic shuttling of viral elements, supported by several host factors, is essential for the replication of the human immunodeficiency virus (HIV). HIV-1 uses a nuclear RNA export pathway mediated by viral protein Rev to transport its Rev response element (RRE)-containing partially spliced and unspliced transcripts aided by the host nuclear RNA export protein CRM1. The factor(s) interacting with the CRM1-Rev complex are potential antiretroviral target(s) and could serve as a retroviral model system to study nuclear export machinery adapted by these viruses. We earlier reported that cellular Staufen-2 interacts with Rev, facilitating viral-RNA export. Here, we identified the formation of a complex between Staufen-2, CRM1 and Rev. Molecular docking and simulations mapped the interacting residues in the RNA-binding Domain 4 of Staufen-2 as R336 and R337, which were experimentally verified to be critical for interactions among Staufen-2, CRM1 and Rev by mutational analysis. Staufen-2 mutants defective in interaction with CRM1 or Rev failed to supplement the Rev-RNA export activity and viral production, demonstrating the importance of these interactions. Rev-dependent reporter assays and proviral DNA-construct transfection-based studies in Staufen-2 knockout cells in the presence of leptomycin-B (LMB) revealed a significant reduction in CRM1-mediated Rev-dependent RNA export with decreased virus production as compared to Staufen-2 knockout background or LMB treatment alone, suggesting the relevance of these interactions in augmenting RNA export activity of Rev. Our observations provide further insights into the mechanistic intricacies of unspliced viral-RNA export to the cytoplasm and support the notion that abrogating such interactions can reduce HIV-1 proliferation.

Abstract

Pattern mining from graph transactional data (GTD) is an active area of research with applications in the domains of bioinformatics, chemical informatics and social networks. Existing works address the problem of mining frequent subgraphs from GTD. However, the knowledge concerning the coverage aspect of a set of subgraphs is also valuable for improving the performance of several applications. In this regard, we introduce the notion of subgraph coverage patterns (SCPs). Given a GTD, a subgraph coverage pattern is a set of subgraphs subject to relative frequency, coverage and overlap constraints provided by the user. We propose the Subgraph ID-based Flat Transactional (SIFT) framework for the efficient extraction of SCPs from a given GTD. Our performance evaluation using three real datasets demonstrates that our proposed SIFT framework is indeed capable of efficiently extracting SCPs from GTD. Furthermore, we demonstrate the effectiveness of SIFT through a case study in computer-aided drug design.

Abstract

Application of deep learning techniques for de novo generation of molecules, termed as inverse molecular design, has been gaining enormous traction in drug design. The representation of molecules in SMILES notation as a string of characters enables the usage of state of the art models in natural language processing, such as Transformers, for molecular design in general. Inspired by generative pre-training (GPT) models that have been shown to be successful in generating meaningful text, we train a transformer-decoder on the next token prediction task using masked self-attention for the generation of druglike molecules in this study. We show that our model, MolGPT, performs on par with other previously proposed modern machine learning frameworks for molecular generation in terms of generating valid, unique, and novel molecules. Furthermore, we demonstrate that the model can be trained conditionally to

Abstract

The discovery of new molecules and materials helps expand the horizons of novel and innovative real-life applications. In pursuit of finding molecules with desired properties, chemists have traditionally relied on experimentation and recently on combinatorial methods to generate new substances often complimented by computational methods. The sheer size of the chemical space makes it infeasible to search through all possible molecules exhaustively. This calls for fast and efficient methods to navigate the chemical space to find substances with desired properties. This class of problems is referred to as inverse design problems. There are a variety of inverse problems in chemistry encompassing various subfields like drug discovery, retrosynthesis, structure identification, etc. Recent developments in modern machine learning (ML) methods have shown great promise in tackling problems of this kind. This has …

Abstract

Drug discovery is a long, expensive, and extremely laborious process that involves multiple steps with knowledge from a wide variety of domains like chemistry, biology and pharmacology. The first step in this process is the identification of potential hit molecules for a novel target followed by experimental evaluation typically using biochemical assays toward lead identification. These hits are then optimized to have higher binding affinity, low toxicity, and improved bioavailability among other requirements. The time and expense involved in this process has given rise to alternate in silico approaches like virtual screening wherein molecules are computationally evaluated to identify potential hits. The structure based drug design (SBDD) method, docking, is used most commonly in virtual screening to identify molecules with high binding affinity to the given target (1–

Abstract

Artificial Intelligence has seen a significant resurgence in the past decade in wide ranging technology and domain areas. Recent progress in digitisation and high influx of biomedical data have led to an unparalleled success of Machine Learning systems in healthcare, which is perceived to be a possible game changer for ‘healthcare to all’. This article gives an account of some of the current applications of AI solutions in the medical domains of diagnosis, prognosis and treatment. The article will also illustrate the implications of AI in the fight against the COVID-19 pandemic. Lastly, the article will summarise the challenges AI currently faces in its wide-scale adoption in the healthcare industry and how they can possibly be dealt with to move towards a more intelligent medical future. This may enable moving towards quality healthcare for all.

Abstract

There has been tremendous advancement in machine learning (ML) applications in computational chemistry, particularly in neural network potentials (NNP). NNPs can approximate potential energy surface (PES) as a high dimensional function by learning from existing reference data, thereby circumventing the need to solve the electronic Schrödinger equation explicitly. As a result, ML accelerates chemical space exploration and property prediction compared to quantum mechanical methods. Novel ML methods have the potential to provide efficient means for predicting the properties of molecules. However, this potential has been limited by the lack of standard comparative evaluations. In this work, we compare four selected models, that is, ANI, PhysNet, SchNet, and BAND‐NN, developed to represent the PES of small organic molecules. We evaluate these models for their accuracy and transferability on two

Abstract

Protein–drug interactions play important roles in many biological processes and therapeutics. Predicting the binding sites of a protein helps to discover such interactions. New drugs can be designed to optimize these interactions, improving protein function. The tertiary structure of a protein decides the binding sites available to the drug molecule, but the determination of the 3D structure is slow and expensive. Conversely, the determination of the amino acid sequence is swift and economical. Although quick and accurate prediction of the binding site using just the sequence is challenging, the application of Deep Learning, which has been hugely successful in several biochemical tasks, makes it feasible. BiRDS is a Residual Neural Network that predicts the protein’s most active binding site using sequence information. SC-PDB, an annotated database of druggable binding sites, is used for training the network …

Abstract

A system and method for exploring a chemical space during molecular design for at least one top hit molecule using a machine learning (ML) model are provided. The method includes (i) representing the at least one molecule stored in a drug library into at least one vector;(ii) clustering the at least one vector to obtain at least one cluster of molecules into one or more clusters;(iii) uniformly sampling a first subset of molecules from each cluster of molecules;(vi) determining a docking score for sampled subset of molecules;(iv) training the ML model by correlating sampled subset of molecules with docking score;(viii) computing acquisition function values for a second subset of molecules from each cluster; and (ix) determining at least one top hit molecule based on the computed acquisition function values, thereby exploring the chemical space for the at least one top hit molecule.