Machine learning
The latest chemistry news and research on machine learning, including reaction prediction, synthesis planning and drug and material discovery, from the Royal Society of Chemistry's magazine, Chemistry World
-
ResearchAI helps chemists consider more options when it comes to picking the best reactants for an experiment
Algorithm can standardise substrate selection to improve performance and structural diversity
-
WebinarEl Agente: ask me anything! A live demonstration
Join us as we tackle your chemistry challenges live with the El Agente team on 24 November
-
NewsGoogle DeepMind’s mutation Atlas shines a light on our dark DNA
AI’s predictions on the impact of changing each and every base of the human genome will help scientists understand genetic basis of disease
-
ResearchBenchmark flaws may be distorting AI drug-discovery leaderboards
Audit identifies subtle but consequential issues, such as conflicting labels and train–test leakage, in numerous benchmark datasets
-
NewsUS National Science Foundation putting $380 million into building ‘self-driving’ labs
North Carolina State is one of the institutes receiving a $20 million grant aims to advance an autonomous chemistry lab
-
FeatureHow AI is transforming the world of analytical chemistry
Analytical chemists have been using machine learning long before ChatGPT made headlines. Now, as generative AI enters the laboratory, the discipline faces both opportunities and risks
-
OpinionCould AI undermine our (belief in the truth of) scientific theories?
Artificial intelligence models that make predictions based solely on data present problems for philosophers
-
FeatureAI agents accelerate catalyst discovery from simulation to scale-up
Artificial intelligence tools are transforming catalyst research, with new AI agents capable of completing in minutes what once took computational chemists days. Andy Extance explores how all scientists can benefit, from small groups to those at tech giants like Meta, Google and Nvidia
-
ResearchStripped down bacterium functions with one less amino acid than the rest of life on Earth
Deep learning models used to understand how isoleucine could be replaced in proteins without disrupting their structures
-
Research‘Human-in-the-loop’ approach could cut costs of robotic labs by 90%
Reducing costs may democratise field by making self-driving labs more accessible
-
OpinionWhy data visualisation is so important in the age of AI
Machines don’t need to draw charts and plots to work with data, but humans will still need datavis skills to work with machines.
-
ResearchMachine learning cuts complexity of computational calculations in catalysis
Software simulates 370,000 steps in under 100 hours, potentially cutting demand for time on supercomputers by orders of magnitude
-
ResearchMolecules with largest dipole moments aren’t due to electronegativity differences
Computational analysis challenges Pauling’s idea of polarity
-
PodcastAlphaGenome & the RNA world hypothesis
In this episode, we discuss Google DeepMind’s latest deep learning model AlphaGenome, dissect the origins of life from chemicals to complex lifeforms, and hear the latest headlines.
-
ResearchGoogle’s AlphaGenome wants to do for DNA what AlphaFold did for proteins
Model predicts effect of mutations on sequences up to 1 million base pairs in length and is adept at tackling complex non-coding regions
-
ResearchAI agents set to democratise computational chemistry
Large language models are powering a new generation of AI agents that could transform computational chemistry from a specialist discipline into one any researcher can use, reports Julia Robinson
-
ResearchProtein design takes big leap forward as model produces enzymes almost as effective as nature’s
New iteration of RoseTTAFold Diffusion optimises side chain placing to break bonds
-
OpinionWhat’s on your lab technology wish list?
Once-magical tools are now mundane. What will the next miracles be?
-
ResearchChemistry and AI pushes biosignature detection for life back billions of years
Combining analytical chemistry with machine learning doubled the age of ancient rocks that can be examined for signs of primordial life
-
ResearchNeural network trained to classify crystal structure errors in MOF and other databases
Study serves as a reminder that machine learning models are only as good as the data they are trained on