Nature Portfolio’s Post

Jump back 23 years into the past and you would have found me growing microscopic protein crystals that I would bombard with X-rays to glimps the atomic level interactions of its binding site. It took me over a year to achieve this to enough resolution with thousands of failed attempts. Fast forward to today and we now have powerful enough computers and AI to achieve what I was doing in a matter of clicks which will drastically speed up drug discovery. Software innovation (with the infrastructure that allows it to run) is the future of life science but it will still need the researchers to provide the inputs and challenge the results. That said, lowly Research Assistants like I was, making crystals, may be in short demand from now on :) https://lnkd.in/eS2NcrHu

Supervised vs. Unsupervised  Learning This framework is a helpful guide for selecting the appropriate algorithm based on the data type and research objective (example for biomedical studies) 🙂

Here, Cai et al. show that brain organoids hooked up to input-output electrodes can be trained to recognize voices and perform other computational tasks! Such "brainoware" may act as a more environmentally friendly form of electronics as well as more easily solve problems that are difficult for classical computers. I will note that we should be cautious to avoid engineering brainoware systems that might be capable of suffering. Nonetheless, this represents an exciting new approach for artificial intelligence! #neurotech #biotech #ai https://lnkd.in/gfcsyW8d

ETH Zürich Chemists unveil #DRAGONFLY, a novel interactome-based deep learning approach for de novo drug design, generating tailored drug-like molecules with desired bioactivity and synthesizability. 🎯 DRAGONFLY (Drug-target interActome-based GeneratiON oF noveL biologicallY active molecules) fuses graph neural networks and chemical language models for the creation of innovative bioactive molecules having applications in bioorganic and medicinal chemistry. Quick Read: https://lnkd.in/gcCztwst #bioinformatics #drugdesign #sbdd #clm #llm #artificialintelligence #machinelearning #cheminformatics #sciencenews #biotechnology

Can Diffusion Models Crack the Protein Structure and Docking Code? 🎯 Diffusion Models have emerged as a powerful new generative AI tool in computational structural biology, delivering major wins in protein design & molecular docking. Explore the potential of Diffusion Models in protein structure and docking. Discover More: https://lnkd.in/d-ukNhT7 #bioinformatics #structuralbiology #ai #diffusionmodels #docking #proteindesign #sciencenews

How Machine Learning algorithms can be used in the biomedical field! And that's not all. There's a scientific research area called Structural Biology, where scientists study the structure of proteins. It is important for all living creatures because the function of a protein depends on its shape. Despite years of relentless effort, we humans have uncovered only a minute fraction of protein structures less than 1% out of 100 to 200 million proteins in total. Then AI comes. In just two years, the AI system called AlphaFold predicted the structures of nearly all proteins in the world, now saved in the open database online. This breakthrough isn't just about proteins, AlphaFold has opened up new avenues for research in various fields, including drug discovery, disease diagnosis, and material sciences. #machinelearning #ai #biomedicalengineering #scientificresearch

Scientists combine AI with cerebral organoids (minibrains) for enhanced computing power #AI #artificialintelligence #biocomputers #biological #Biotechnology #cerebralorganoids #computingpower #energyefficientalternatives #hybridalgorithm #llm #machinelearning #mathematicaltasks #minibrains #neurodegenerativeconditions #ReinforcementLearning #reservoircomputing #revolutionizedrugtesting #SpeechRecognition #technological

The signal variability that is present in some multiplex immunofluorescence (mIF) methods can limit the adoption of this highly sensitive technology. By using the power of AI to strengthen image analysis, Lorenz Rognoni, PhD, Spatial Image Analysis Portfolio Owner at Ultivue, is hoping to make mIF a primary endpoint assay in clinical trials. The Image Data Science and AI Engineering teams at Ultivue are training deep learning models with more than signal intensity thresholds. Learn how the addition of cellular morphology and contextual details into the analysis enables more accurate identification of positive and negative cellular phenotypes: https://hubs.ly/Q02crf6L0 #TumorMicroenvironment #MultiplexImmunofluorescence #biomarkers #ImageAnalysis #Spatial Biology

A new deep-learning-based analysis toolkit for spatial transcriptomics Spatial transcriptomics (ST) technologies are applied in biology and medical research for its ability to detect the spatial distribution of transcriptome in histological tissue slices. By probing some of the transcripts or performing sequencing, the researchers are able to know about the transcription level in cells. Then they can further predict the cell types and build the three-dimensional (3D) structure of the tissue according to the information. However, the analysis could be difficult when there are multiple slices need to be analyze jointly using state of the art toolkits. It will bring challenge for researchers to manually assemble the slices and build the 3D structure. To overcome the problem, a research team developed a new spatial architecture characterization by deep learning (SPACEL). Through three modules, Spoint, Splane and Scube, SPACEL can build the 3D panorama of tissues automatically. SPACEL has demonstrated its superior performance over the others for cell type deconvolution in three core analytical tasks: predicting cell type distribution, identifying spatial domains, and reconstructing three-dimensional tissue structures. #ScienceMission #sciencenewshighlights https://lnkd.in/gG7iGGXm

Hello connections, I am thrilled to inform you that our research paper titled "HivNet: Studying in depth the morphology of HIV-1 virion using Deep Learning" has been accepted and published in Frontiers in Biomedical Technologies Journal (FBT) journal this year. Our work categorises the three stages of HIV-1 virions maturity and morphogenesis with the devoted application of deep learning techniques. We anticipate that this tool will be useful to a variety of studies. DOI - https://lnkd.in/gZxMKJCy