2026-10-12T08:19:47Z https://www.scidb.cn/oai

10.57760/sciencedb.012sk 2026-09-20T10:40:44Z

2026-09-20 ScaProp-Instructions: A Multimodal Large Language Model scaffold Constrained Molecular Generation Dataset doi:10.57760/sciencedb.012sk en ScaProp-Instructions is a multimodal instruction dataset developed for scaffold-constrained and property-controllable molecular generation. The dataset is constructed based on the MolOpt-Instructions dataset and is designed to establish a unified mapping between molecular scaffold structures, natural-language property constraints, and target molecular structures.The dataset construction process consists of several stages. First, molecular SMILES sequence pairs are obtained from MolOpt-Instructions and filtered according to chemical validity. The Bemis--Murcko scaffold extraction algorithm is then applied to the molecular pairs to identify their core structural frameworks. The resulting molecular scaffolds are represented as standardized two-dimensional scaffold images, which serve as explicit structural constraints during multimodal molecular generation.In parallel, the changes in molecular physicochemical properties associated with molecular modification are quantitatively analyzed, including molecular weight (MW) and logarithm of the partition coefficient (LogP). Based on these property changes, together with predefined prompt templates, natural-language instructions are reconstructed to describe the desired property constraints for molecular generation.The resulting samples are organized as multimodal triplets consisting of: (1) a molecular scaffold image, (2) a natural-language property instruction, and (3) a target molecular structure represented by SMILES. These triplets provide aligned visual, textual, and molecular information for training and evaluating multimodal large language models in scaffold-preserving molecular generation.The dataset is subsequently divided into training, validation, and test sets according to a predefined ratio. The test set is constructed to contain independent molecular samples with non-overlapping molecular skeleton topologies relative to the training and validation sets, enabling evaluation of model generalization to unseen chemical scaffolds.ScaProp-Instructions is intended to support research on multimodal molecular generation, scaffold preservation, property-controlled molecular optimization, and vision-language alignment in the chemical domain. molecular generation; molecular scaffold; scaffold constraint; property control; multimodal molecular generation; molecular images; chemical language models; drug design; lead optimization Yang Shuting Yu Zihao Cheng Debo Huang Yu Feng Zaiwen Li Chen RESTRICTED dataset Science Data Bank