Collaborating Authors


Top 5 Machine learning models 2021


This year has been full of a lot of great models. In this article, my hope is to highlight 10 of the most noteworthy models. I have been regularly reviewing papers and explaining them over this year and I think I have quite a few good mentions. Disclaimer: There might be other good models not mentioned here and I am not claiming to be the ultimate expert when it comes to evaluating the quality of machine learning models! Also, note that this list isn't ordered!

A proteomic survival predictor for COVID-19 patients in intensive care


Author summary Healthcare systems around the world are struggling to accommodate high numbers of the most severely ill patients with COVID-19. Moreover, the pandemic creates a pressing need to accelerate clinical trials investigating potential new therapeutics. While various biomarkers can discriminate and predict the future course of disease for patients of different disease severity, prognosis remains difficult for patient groups with similar disease severity, e.g. patients requiring intensive care. Established risk assessments in intensive care medicine such as the SOFA or APACHE II show only limited reliability in predicting future disease outcomes for COVID-19. In this study we hypothesized that the plasma proteome, which reflects the complete set of proteins that are expressed by an organism and are present in the blood, and which is known to comprehensively capture the host response to COVID-19, can be leveraged to allow for prediction of survival in the most critically ill patients with COVID-19. Here, we found 14 proteins, which over time changed in opposite directions for patients who survive compared to patients who do not survive on intensive care. Using a machine learning model which combines the measurements of multiple proteins, we were able to accurately predict survival in critically ill patients with COVID-19 from single blood samples, weeks before the outcome, substantially outperforming established risk predictors.

Deep Learning Based DNA and RNA Binding Sites Prediction for Accelerated Drug Discovery - CBIRT


Scientists from Skoltech's iMolecule group have created an artificial intelligence-driven approach to identify sites on the structures of DNA or RNA molecules where drug compounds may bind. The drug-binding site information will allow pharmaceutical firms to find novel therapeutic compounds – including antiviral agents – in a far more focused manner. The new method, which was published in Nucleic Acid Research: Genomics and Bioinformatics, claims to be more accurate than previous methods since it considers how a nucleic acid molecule's shape impacts which binding sites are accessible. Most drugs target proteins because pharmacologists have traditionally seen RNA as just a mediator between DNA and the functional proteins it encodes. As almost 85% of the genome is translated into RNAs, only a tiny percentage of those RNAs encode proteins.

Researchers Use Machine Learning To Repair Genetic Damage


DNA damage is constantly occurring in cells, either due to external sources or as a result of internal cellular metabolic reactions and physiological activities. Accurate repair of such DNA damages is critical to avoid mutations and chromosomal rearrangements linked to diseases including cancer, immunodeficiencies, neurodegeneration, and premature aging. A team of researchers at Massachusetts General Hospital and the National Cancer Research Centre have identified a way to repair genetic damage and prevent DNA alterations using machine learning techniques. The researchers state that it is possible to learn more about how cancer develops and how to fight it if we understand how DNA lesions originate and repair. Therefore, they hope that their discovery will help create better cancer treatments while also protecting our healthy cells. To combat challenges to DNA integrity, cells have evolved systems that detect DNA lesions and initiate a signaling cascade that promotes DNA repair, referred to as the DNA damage response (DDR).

Absci and deep learning's quest for the perfect protein


The breakthrough of CRISPR technology in the past two decades has allowed biologists to refine the manipulation of DNA, to slice and dice it in order to create organisms tailored to particular purposes. That free-wheeling editing of genes, though, produces a new problem: how to organize all the complexity of the different edited pieces of DNA. That's especially important for the multi-hundred-billion-dollar portion of the drug market called biologics, basically engineered proteins that can achieve a particular purpose. If you're going to engineer new proteins through CRISPR, you need to do it in a systemic way, which is a fairly demanding combinatorial problem. Hence, some smart young biotechs are turning to deep learning forms of artificial intelligence, as deep learning is a technology that loves combinatorial problems.

Artificial intelligence in 2021: the AIhub roundup


As 2021 draws to a close, we look back on some of the AI research, news, policy developments and awards that have piqued our interest. We start our round-up with awards, with many prestigious prizes being presented this year. Launched last year, the AAAI Squirrel AI Award recognises positive impacts of artificial intelligence to protect, enhance, and improve human life in meaningful ways with long-lived effects. The winner of the 2021 award was Cynthia Rudin for work in interpretable and transparent AI systems in real-world deployments. The 2021/2022 ACM Athena Lecturer Award went to Ayanna Howard, who was recognised for fundamental contributions to the development of accessible human-robotic systems and AI, and for her efforts to broaden participation in computing through entrepreneurial and mentoring.

Predicting Death Could Change the Value of a Life


If you could predict your death, would you want to? For most of human history, the answer has been a qualified yes. In Neolithic China, seers practiced pyro-osteomancy, or the reading of bones; ancient Greeks divined the future by the flight of birds; Mesopotamians even attempted to plot the future in the attenuated entrails of dead animals. We've looked to the stars and the movement of planets, we've looked to weather patterns, and we've even looked to bodily divinations like the "child born with a caul" superstition to assure future good fortune and long life. By the 1700s, the art of prediction had grown slightly more scientific, with mathematician and probability expert Abraham de Moivre attempting to calculate his own death by equation, but truly accurate predictions remained out of reach.

A JUST Food System Starts with Breakfast


JUST is searching faster and further, collaborating with rural farmers, Michelin-starred chefs, and world-class product developers to discover more effective tools within the plant kingdom to make food tastier, healthier, and more sustainable. But it all starts with the tools. And there happens to be 400,000 of them. Collectively, these species possess 18 billion proteins, 108 million fats, and 4 million carbohydrates. Most of them have never even been explored for the potential to make our cookies, pasta, ice cream, butter, or scrambled eggs better.

AI In Biotechnology Causing to Revamp Pharmaceuticals


Biotechnology in itself is an extremely diverse field that engages itself in various exploration concerning the living organisms on the surface of the planet. When artificial intelligence is combined with biotechnology, the mundane process is evaded by the advanced applications empowered by artificial intelligence. Healthtech derives immense benefits from the applications of biotechnology that are entitled to pharmaceuticals. Thus, AI in biotechnology is expanding the opportunities awaiting the pharmaceuticals industry. Artificial intelligence has facilitated an advanced level of research in the field of biotechnology by improving the quality by 10 times.

De Novo Structure-Based Drug Design Using Deep Learning


In recent years, deep learning-based methods have emerged as promising tools for de novo drug design. Most of these methods are ligand-based, where an initial target-specific ligand data set is necessary to design potent molecules with optimized properties. Although there have been attempts to develop alternative ways to design target-specific ligand data sets, availability of such data sets remains a challenge while designing molecules against novel target proteins. In this work, we propose a deep learning-based method, where the knowledge of the active site structure of the target protein is sufficient to design new molecules. First, a graph attention model was used to learn the structure and features of the amino acids in the active site of proteins that are experimentally known to form protein–ligand complexes.