Hartog et al. investigate ways to speed up the Chemformer retrosynthesis model with knowledge distillation, alternate architectures, and hyper-parameter tuning. pubs.acs.org/doi/full/10....
Investigations into the Efficiency of Computer-Aided Synthesis Planning
The efficiency of machine learning (ML) models is crucial to minimize inference times and reduce the carbon footprints of models deployed in production environments. Current models employed in retrosy...
pubs.acs.org
February 2, 2025 at 6:06 PM
Hartog et al. investigate ways to speed up the Chemformer retrosynthesis model with knowledge distillation, alternate architectures, and hyper-parameter tuning. pubs.acs.org/doi/full/10....
I recently gave a talk at ML in PL about our recent progress on retrosynthesis prediction; the recording is now out:
youtube.com/watch?v=hfUL...
youtube.com/watch?v=hfUL...
youtube.com
February 5, 2025 at 7:07 PM
I recently gave a talk at ML in PL about our recent progress on retrosynthesis prediction; the recording is now out:
youtube.com/watch?v=hfUL...
youtube.com/watch?v=hfUL...
Sean Current, Ziqi Chen, Daniel Adu-Ampratwum, Xia Ning, Srinivasan Parthasarathy: DiffER: Categorical Diffusion for Chemical Retrosynthesis https://arxiv.org/abs/2505.23721 https://arxiv.org/pdf/2505.23721 https://arxiv.org/html/2505.23721
May 30, 2025 at 6:25 AM
Sean Current, Ziqi Chen, Daniel Adu-Ampratwum, Xia Ning, Srinivasan Parthasarathy: DiffER: Categorical Diffusion for Chemical Retrosynthesis https://arxiv.org/abs/2505.23721 https://arxiv.org/pdf/2505.23721 https://arxiv.org/html/2505.23721
Zhang, Liu, Zhang, Xiong, Zhai, Hao, Gu, Yang, Gao, Hu, Zhou, He: A Large Language Model for Chemistry and Retrosynthesis Predictions https://arxiv.org/abs/2507.01444 https://arxiv.org/pdf/2507.01444 https://arxiv.org/html/2507.01444
July 3, 2025 at 6:47 AM
Zhang, Liu, Zhang, Xiong, Zhai, Hao, Gu, Yang, Gao, Hu, Zhou, He: A Large Language Model for Chemistry and Retrosynthesis Predictions https://arxiv.org/abs/2507.01444 https://arxiv.org/pdf/2507.01444 https://arxiv.org/html/2507.01444
𝗥𝗲𝗮𝗰𝘁𝗶𝗼𝗻 𝗣𝗹𝗮𝗻𝗻𝗶𝗻𝗴 𝘄𝗶𝘁𝗵 𝗗𝗲𝗲𝗽 𝗟𝗲𝗮𝗿𝗻𝗶𝗻𝗴 ⚛️:
➡️ Machine learning frameworks for chemical retrosynthesis: pubs.rsc.org/en/content/a...
➡️ Machine learning frameworks for chemical retrosynthesis: pubs.rsc.org/en/content/a...
Investigating the reliability and interpretability of machine learning frameworks for chemical retrosynthesis
Machine learning models for chemical retrosynthesis have attracted substantial interest in recent years. Unaddressed challenges, particularly the absence of robust evaluation metrics for performance c...
pubs.rsc.org
January 14, 2025 at 6:49 PM
𝗥𝗲𝗮𝗰𝘁𝗶𝗼𝗻 𝗣𝗹𝗮𝗻𝗻𝗶𝗻𝗴 𝘄𝗶𝘁𝗵 𝗗𝗲𝗲𝗽 𝗟𝗲𝗮𝗿𝗻𝗶𝗻𝗴 ⚛️:
➡️ Machine learning frameworks for chemical retrosynthesis: pubs.rsc.org/en/content/a...
➡️ Machine learning frameworks for chemical retrosynthesis: pubs.rsc.org/en/content/a...
Mianchu Wang, Giovanni Montana: Retrosynthesis Planning via Worst-path Policy Optimisation in Tree-structured MDPs https://arxiv.org/abs/2509.10504 https://arxiv.org/pdf/2509.10504 https://arxiv.org/html/2509.10504
September 16, 2025 at 6:32 AM
Mianchu Wang, Giovanni Montana: Retrosynthesis Planning via Worst-path Policy Optimisation in Tree-structured MDPs https://arxiv.org/abs/2509.10504 https://arxiv.org/pdf/2509.10504 https://arxiv.org/html/2509.10504
Higher-level Strategies for Computer-Aided Retrosynthesis
Authors: Jihye Roh, Joonyoung F. Joung, Kevin Yu, Zhengkai Tu, G. Logan Bartholomew, Omar A. Santiago-Reyes, Mun Hong Fong, Richmond Sarpong, Sarah E. Reisman, Connor W. Coley
DOI: 10.26434/chemrxiv-2025-21zvt
Authors: Jihye Roh, Joonyoung F. Joung, Kevin Yu, Zhengkai Tu, G. Logan Bartholomew, Omar A. Santiago-Reyes, Mun Hong Fong, Richmond Sarpong, Sarah E. Reisman, Connor W. Coley
DOI: 10.26434/chemrxiv-2025-21zvt
February 5, 2025 at 11:00 AM
Higher-level Strategies for Computer-Aided Retrosynthesis
Authors: Jihye Roh, Joonyoung F. Joung, Kevin Yu, Zhengkai Tu, G. Logan Bartholomew, Omar A. Santiago-Reyes, Mun Hong Fong, Richmond Sarpong, Sarah E. Reisman, Connor W. Coley
DOI: 10.26434/chemrxiv-2025-21zvt
Authors: Jihye Roh, Joonyoung F. Joung, Kevin Yu, Zhengkai Tu, G. Logan Bartholomew, Omar A. Santiago-Reyes, Mun Hong Fong, Richmond Sarpong, Sarah E. Reisman, Connor W. Coley
DOI: 10.26434/chemrxiv-2025-21zvt
optimize the process of recombination and fragmentation as well as the tasks between reaction and retrosynthesis prediction. Extensive experiments on Mol-Instruction and USPTO-50K datasets demonstrate that ChemDual achieves state-of-the-art [5/7 of https://arxiv.org/abs/2505.02639v1]
May 6, 2025 at 6:16 AM
optimize the process of recombination and fragmentation as well as the tasks between reaction and retrosynthesis prediction. Extensive experiments on Mol-Instruction and USPTO-50K datasets demonstrate that ChemDual achieves state-of-the-art [5/7 of https://arxiv.org/abs/2505.02639v1]
I wonder what its retrosynthesis would be - I'd like to know where to get multivalent hydrogen.
October 8, 2025 at 7:19 PM
I wonder what its retrosynthesis would be - I'd like to know where to get multivalent hydrogen.
the capability of LLMs, particularly the Llama-3.1 family (8B and 70B), across three core biochemical tasks: Enzyme Commission number prediction, forward synthesis, and retrosynthesis. We compare single-task and multitask learning strategies, [3/6 of https://arxiv.org/abs/2505.05616v1]
May 12, 2025 at 5:54 AM
the capability of LLMs, particularly the Llama-3.1 family (8B and 70B), across three core biochemical tasks: Enzyme Commission number prediction, forward synthesis, and retrosynthesis. We compare single-task and multitask learning strategies, [3/6 of https://arxiv.org/abs/2505.05616v1]
these tasks faces two major challenges: (i) lacking a large-scale chemical synthesis-related instruction dataset; (ii) ignoring the close correlation between reaction and retrosynthesis prediction for the existing fine-tuning strategies. To address [2/7 of https://arxiv.org/abs/2505.02639v1]
May 6, 2025 at 6:16 AM
these tasks faces two major challenges: (i) lacking a large-scale chemical synthesis-related instruction dataset; (ii) ignoring the close correlation between reaction and retrosynthesis prediction for the existing fine-tuning strategies. To address [2/7 of https://arxiv.org/abs/2505.02639v1]
Retrieval‑Retro, an AI system for inorganic retrosynthesis with source code on GitHub, outperforms baseline models by identifying more novel and chemically plausible synthesis routes. https://getnews.me/retrieval-retro-ai-driven-inorganic-retrosynthesis/ #inorganicchemistry #retrosynthesis #ai
September 20, 2025 at 7:35 AM
Retrieval‑Retro, an AI system for inorganic retrosynthesis with source code on GitHub, outperforms baseline models by identifying more novel and chemically plausible synthesis routes. https://getnews.me/retrieval-retro-ai-driven-inorganic-retrosynthesis/ #inorganicchemistry #retrosynthesis #ai
Xuan Lin, Qingrui Liu, Hongxin Xiang, Daojian Zeng, Xiangxiang Zeng: Enhancing Chemical Reaction and Retrosynthesis Prediction with Large Language Model and Dual-task Learning https://arxiv.org/abs/2505.02639 https://arxiv.org/pdf/2505.02639 https://arxiv.org/html/2505.02639
May 6, 2025 at 6:16 AM
Xuan Lin, Qingrui Liu, Hongxin Xiang, Daojian Zeng, Xiangxiang Zeng: Enhancing Chemical Reaction and Retrosynthesis Prediction with Large Language Model and Dual-task Learning https://arxiv.org/abs/2505.02639 https://arxiv.org/pdf/2505.02639 https://arxiv.org/html/2505.02639
knowledge, with multitask learning enhancing forward- and retrosynthesis predictions by leveraging shared enzymatic information. We also identify key limitations, for example challenges in hierarchical EC classification schemes, highlighting areas for [5/6 of https://arxiv.org/abs/2505.05616v1]
May 12, 2025 at 5:54 AM
knowledge, with multitask learning enhancing forward- and retrosynthesis predictions by leveraging shared enzymatic information. We also identify key limitations, for example challenges in hierarchical EC classification schemes, highlighting areas for [5/6 of https://arxiv.org/abs/2505.05616v1]
Chemma: Accelerating organic synthesis with Large Language Models (LLMs). Yu Zhang and coauthors (Nature Machine Intelligence) present Chemma, a fine-tuned LLM excelling in retrosynthesis and yield prediction, optimizing unexplored reactions rapidly. www.nature.com/articles/s42...
Large language models to accelerate organic chemistry synthesis - Nature Machine Intelligence
Large language models (LLMs) can be useful tools for science, but they often lack expert understanding of complex domains that they were not trained on. Zhang and colleagues fine-tuned a LLaMA-2-7b-ba...
www.nature.com
July 6, 2025 at 8:15 PM
Chemma: Accelerating organic synthesis with Large Language Models (LLMs). Yu Zhang and coauthors (Nature Machine Intelligence) present Chemma, a fine-tuned LLM excelling in retrosynthesis and yield prediction, optimizing unexplored reactions rapidly. www.nature.com/articles/s42...
Our retrosynthesis model RetroChimera (previously Chimera) is available via Azure Foundry. Its prediction quality has now surpassed the patent data itself (i.e. in blind tests chemists prefer model outputs over real published reactions). Exciting times ahead 🔥
RetroChimera, now available on Azure AI Foundry, marks a new milestone for predicting synthesis routes to drug-like molecules, opening new possibilities for AI in drug discovery. Learn more: msft.it/6011sutzJ
August 13, 2025 at 6:06 PM
Our retrosynthesis model RetroChimera (previously Chimera) is available via Azure Foundry. Its prediction quality has now surpassed the patent data itself (i.e. in blind tests chemists prefer model outputs over real published reactions). Exciting times ahead 🔥
We just preprinted Chimera: our new retrosynthesis prediction model ⚗️
aka.ms/chimera
It has two modules: a large-scale Transformer and a graph-rewriting GNN. Chimera learns to combine predictions from both to leverage their unique strengths & biases.
Thread below 🧵
aka.ms/chimera
It has two modules: a large-scale Transformer and a graph-rewriting GNN. Chimera learns to combine predictions from both to leverage their unique strengths & biases.
Thread below 🧵
December 11, 2024 at 9:30 PM
We just preprinted Chimera: our new retrosynthesis prediction model ⚗️
aka.ms/chimera
It has two modules: a large-scale Transformer and a graph-rewriting GNN. Chimera learns to combine predictions from both to leverage their unique strengths & biases.
Thread below 🧵
aka.ms/chimera
It has two modules: a large-scale Transformer and a graph-rewriting GNN. Chimera learns to combine predictions from both to leverage their unique strengths & biases.
Thread below 🧵
Alan Kai Hassen, Andrius Bernatavicius, Antonius P. A. Janssen, Mike Preuss, Gerard J. P. van Westen, Djork-Arn\'e Clevert: Atom-anchored LLMs speak Chemistry: A Retrosynthesis Demonstration https://arxiv.org/abs/2510.16590 https://arxiv.org/pdf/2510.16590 https://arxiv.org/html/2510.16590
October 21, 2025 at 6:34 AM
Alan Kai Hassen, Andrius Bernatavicius, Antonius P. A. Janssen, Mike Preuss, Gerard J. P. van Westen, Djork-Arn\'e Clevert: Atom-anchored LLMs speak Chemistry: A Retrosynthesis Demonstration https://arxiv.org/abs/2510.16590 https://arxiv.org/pdf/2510.16590 https://arxiv.org/html/2510.16590
machine learning-based retrosynthesis gathers reaction data from multiple sources into one single edge to train prediction models. This paradigm poses considerable privacy risks as it necessitates broad data availability [3/9 of https://arxiv.org/abs/2502.19119v1]
February 27, 2025 at 6:10 AM
machine learning-based retrosynthesis gathers reaction data from multiple sources into one single edge to train prediction models. This paradigm poses considerable privacy risks as it necessitates broad data availability [3/9 of https://arxiv.org/abs/2502.19119v1]
Guikun Chen, Xu Zhang, Yi Yang, Wenguan Wang: Chemical knowledge-informed framework for privacy-aware retrosynthesis learning https://arxiv.org/abs/2502.19119 https://arxiv.org/pdf/2502.19119 https://arxiv.org/html/2502.19119
February 27, 2025 at 6:10 AM
Guikun Chen, Xu Zhang, Yi Yang, Wenguan Wang: Chemical knowledge-informed framework for privacy-aware retrosynthesis learning https://arxiv.org/abs/2502.19119 https://arxiv.org/pdf/2502.19119 https://arxiv.org/html/2502.19119
Chimera uses an ensemble of models with diverse inductive biases, coupled with a learned reranking strategy, to improve retrosynthesis prediction (Figure 1, pg. 1-3). This approach leverages the strengths of different model types. For instance, edit-based models excel with common reaction...
December 9, 2024 at 4:11 PM
Chimera uses an ensemble of models with diverse inductive biases, coupled with a learned reranking strategy, to improve retrosynthesis prediction (Figure 1, pg. 1-3). This approach leverages the strengths of different model types. For instance, edit-based models excel with common reaction...
Transforming retrosynthesis with AI/ML
🗂️: Chemical AI, Chemify, DeepMatter, Pending AI, Iktos, MoleculeOne, Elix, Synthia by Merck and ChemDraw by PerkinElmer (Revvity Signals), Optibrium and many more
open.substack.com/pub/marinata...
#science #health #drugdevelopment #drugdiscovery #AI
#biotech
🗂️: Chemical AI, Chemify, DeepMatter, Pending AI, Iktos, MoleculeOne, Elix, Synthia by Merck and ChemDraw by PerkinElmer (Revvity Signals), Optibrium and many more
open.substack.com/pub/marinata...
#science #health #drugdevelopment #drugdiscovery #AI
#biotech
Transforming retrosynthesis with AI/ML
AI/ML tools for planning and execution of chemical synthesis
open.substack.com
December 3, 2024 at 2:59 PM
Transforming retrosynthesis with AI/ML
🗂️: Chemical AI, Chemify, DeepMatter, Pending AI, Iktos, MoleculeOne, Elix, Synthia by Merck and ChemDraw by PerkinElmer (Revvity Signals), Optibrium and many more
open.substack.com/pub/marinata...
#science #health #drugdevelopment #drugdiscovery #AI
#biotech
🗂️: Chemical AI, Chemify, DeepMatter, Pending AI, Iktos, MoleculeOne, Elix, Synthia by Merck and ChemDraw by PerkinElmer (Revvity Signals), Optibrium and many more
open.substack.com/pub/marinata...
#science #health #drugdevelopment #drugdiscovery #AI
#biotech
Tiffadelic announces debut album, Retrosynthesis. Out September 18.
@synthsky.bsky.social #goth #synthwave #minimalwave #electronicmusic #newmusic
@synthsky.bsky.social #goth #synthwave #minimalwave #electronicmusic #newmusic
August 14, 2025 at 12:07 AM
Tiffadelic announces debut album, Retrosynthesis. Out September 18.
@synthsky.bsky.social #goth #synthwave #minimalwave #electronicmusic #newmusic
@synthsky.bsky.social #goth #synthwave #minimalwave #electronicmusic #newmusic
🔗 A data-driven group retrosynthesis planning model inspired by neurosymbolic programming
#NeuroSymbolic #AI #FutureOfAI
www.nature.com/articles/s41...
#NeuroSymbolic #AI #FutureOfAI
www.nature.com/articles/s41...
A data-driven group retrosynthesis planning model inspired by neurosymbolic programming - Nature Communications
Generative models rapidly generate many similar molecules but face synthesis challenges. Here, authors introduce an algorithm that leverages reusable synthesis patterns to improve retrosynthetic plann...
www.nature.com
January 4, 2025 at 3:19 PM
🔗 A data-driven group retrosynthesis planning model inspired by neurosymbolic programming
#NeuroSymbolic #AI #FutureOfAI
www.nature.com/articles/s41...
#NeuroSymbolic #AI #FutureOfAI
www.nature.com/articles/s41...