Code: Paper: Co-authors: @daniil_boiko, Nikita Kolomoets, @AnanikovLab.(6/6).

Analyzing total syntheses, we found human synthetic strategies differ notably from natural biosynthesis by analyzing strychnine molecules. Modern methods prioritize atom economy, whereas nature uses flexible “ascent–descent” strategies to diversify molecules. (5/6)

Beyond drug molecules, our model helps visualize chemical reaction landscapes by analyzing the reaction atlas developed by @pschwllr. By analyzing the complexity changes, we quantified which reactions classes are critical for efficient synthetic planning. (4/6)

We applied our complexity model to analyze FDA-approved drugs (1985-2022). The clear trend: drugs are becoming structurally more complex, reflecting advances in synthetic and drug development methods. (3/6)

Molecular complexity shapes drug discovery, organic synthesis, and more, but quantifying it has remained subjective. We approached this by training an ML ranking model on nearly 200k expert-labeled data points collected during separate stages. (2/6)

Excited to share our work published in @ChemicalScience, where we digitize molecular complexity — a fundamental but elusive concept in chemistry — by formulating it as a ranking problem!.(1/6)

the ones that come to mind are.- organic chemistry by Jonathan Clayden.- probabilistic machine learning (2 parts) by @sirbayes .- intro to the theory of computation by Michael Sipser.

carefully written textbooks are underappreciated in the modern world.

RT @CorinWagen: Over the past year, I've been learning about how quantum chemistry is used outside academia. (Turns out it's not just for "….