Papers › BioT5+: Towards Generalized Biological Understanding with IUPAC Integration and...

BioT5+: Towards Generalized Biological Understanding with IUPAC Integration and Multi-task Tuning

27 Feb 2024arXiv:2402.17810archive 2025-07-28

Qizhi Pei, Lijun Wu, Kaiyuan Gao, Xiaozhuan Liang, Yin Fang, Jinhua Zhu, Shufang Xie, Tao Qin, Rui Yan

Recent research trends in computational biology have increasingly focused on integrating text and bio-entity modeling, especially in the context of molecules and proteins. However, previous efforts like BioT5 faced challenges in generalizing across diverse tasks and lacked a nuanced understanding of molecular structures, particularly in their textual representations (e.g., IUPAC). This paper introduces BioT5+, an extension of the BioT5 framework, tailored to enhance biological research and drug discovery. BioT5+ incorporates several novel features: integration of IUPAC names for molecular understanding, inclusion of extensive bio-text and molecule data from sources like bioRxiv and PubChem, the multi-task instruction tuning for generality across tasks, and a numerical tokenization technique for improved processing of numerical data. These enhancements allow BioT5+ to bridge the gap between molecular representations and their textual descriptions, providing a more holistic understanding of biological entities, and largely improving the grounded reasoning of bio-text and bio-sequences. The model is pre-trained and fine-tuned with a large number of experiments, including \emph{3 types of problems (classification, regression, generation), 15 kinds of tasks, and 21 total benchmark datasets}, demonstrating the remarkable performance and state-of-the-art results in most cases. BioT5+ stands out for its ability to capture intricate relationships in biological data, thereby contributing significantly to bioinformatics and computational biology. Our code is available at \url{https://github.com/QizhiPei/BioT5}.

PaperPDFCode

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Code

QizhiPei/BioT5 officialmentioned in papermentioned on GitHubpytorch report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Drug DiscoveryForward reaction predictionMolecule CaptioningReagent PredictionRetrosynthesisText-based de novo Molecule Generation

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Forward reaction prediction Mol-Instruction BioT5+ Exact 0.864 #2 of 2 Archive leaderboard report
Forward reaction prediction Mol-Instruction BioT5+ Morgan FTS 0.935 #2 of 2 Archive leaderboard report
Forward reaction prediction Mol-Instruction BioT5+ Validity 1 #2 of 2 Archive leaderboard report
Molecule Captioning ChEBI-20 BioT5+ BLEU-2 66.6 #4 of 33 Archive leaderboard report
Molecule Captioning ChEBI-20 BioT5+ BLEU-4 59.1 #4 of 33 Archive leaderboard report
Molecule Captioning ChEBI-20 BioT5+ METEOR 68.1 #4 of 33 Archive leaderboard report
Molecule Captioning ChEBI-20 BioT5+ ROUGE-1 71.0 #4 of 33 Archive leaderboard report
Molecule Captioning ChEBI-20 BioT5+ ROUGE-2 58.4 #4 of 33 Archive leaderboard report
Molecule Captioning ChEBI-20 BioT5+ ROUGE-L 65.0 #4 of 33 Archive leaderboard report
Reagent Prediction Mol-Instruction BioT5+ Exact 0.257 #2 of 2 Archive leaderboard report
Reagent Prediction Mol-Instruction BioT5+ Morgan FTS 0.512 #2 of 2 Archive leaderboard report
Reagent Prediction Mol-Instruction BioT5+ Validity 1 #2 of 2 Archive leaderboard report
Retrosynthesis Mol-Instruction BioT5+ Exact 0.642 #2 of 2 Archive leaderboard report
Retrosynthesis Mol-Instruction BioT5+ Morgan FTS 0.866 #2 of 2 Archive leaderboard report
Retrosynthesis Mol-Instruction BioT5+ Validity 1 #2 of 2 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 BioT5+ BLEU 87.2 #3 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 BioT5+ Exact Match 52.2 #3 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 BioT5+ Frechet ChemNet Distance (FCD) 0.353 #3 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 BioT5+ Levenshtein 12.776 #3 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 BioT5+ MACCS FTS 90.7 #3 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 BioT5+ Morgan FTS 77.9 #3 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 BioT5+ Parameter Count 252000000 #3 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 BioT5+ RDK FTS 83.5 #3 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 BioT5+ Text2Mol 57.9 #3 of 20 Archive leaderboard report
Text-based de novo Molecule Generation ChEBI-20 BioT5+ Validity 100 #3 of 20 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections