Simulation and Experimental Validation Data for the Discovery of Amyloid-Like Short Peptide Assembly Codes

This dataset supports the manuscript "Discovery of Amyloid-Like Short Peptide Assembly Codes via an Experimentally Validated Predictive Computational Platform" (Zheng, Huang, Lan, Tian, Lee & Han). It provides the coarse-grained/hybrid-resolution molecular dynamics (PACE force field) simulation results and the experimental characterization data used to discover and validate short (3-5 residue) amyloid-forming peptides. The fine-tuned Uni-Mol machine learning model (weights and training data) is maintained together with the analysis code on GitHub (see Related resources below) rather than duplicated here. Contents: 01_simulation_results.zip — Aggregation propensity (AP) and Amyloid-Like Tendency (ALT) scores from exhaustive/extended PACE simulations of tripeptides (n=1,303), tetrapeptides (n=235), and pentapeptides (n=119), together with the final-frame PDB structures of the simulated assemblies and the figure-level source data (SourceData/) for all main and supplementary figures. 02_synthesis_qc.zip — Certificates of Analysis (purity/TFA content) for the 61 tri-, tetra-, and pentapeptides custom-synthesized for experimental validation (20 tripeptides, 26 tetrapeptides, 15 pentapeptides). 03_experimental_analysis.zip — Python/Jupyter scripts used to (i) deconvolute the Amide I region of ATR-FTIR spectra into secondary-structure components and (ii) classify peptides into amyloid-forming/non-amyloid groups via spectral clustering of ThT fluorescence enhancement and FTIR beta-sheet content, along with the compiled experimental summary tables and one worked example. Related resources: Analysis/tutorial code (PACE MD automation, ML training/inference): https://github.com/hanlab-computChem/hanlab Raw SEM, XRD, and full FTIR/ThT spectral datasets for all characterized peptides are available from the corresponding authors upon reasonable request. Please cite the associated manuscript when using this dataset.