生物制药Surflex平台5.191多操作系统

The Surflex Platform consists of the five modules described below. The Surflex Manual contains details of all computational procedures and options within each command-line module. We support Linux (most common variants), Windows, and MacOS. All of the modules are multi-core capable, and very substantial speed-ups are observed with modern multi-core laptops, workstations, and HPC clusters.

Tools Module Fast and Accurate Small Molecule Processing

The Tools module addresses the most common aspects of small-molecule preparation 2D to 3D conversion (from SMILES or SDF) Chirality detection and enumeration Protonation Conformer generation

Features and benefits Template-free and non-stochastic Relies on MMFF94sf forcefield for structure derivation Fast and accurate on typical drug-like ligands, with better coverage of diverse conformations Fastest and most accurate method for macrocyclic ligands Capable of incorporating NMR restraints, which is particularly useful for large peptidic macrocycles

Similarity Module State-of-the-Art 3D Molecular Similarity

The Similarity module implements ligand similarity operations using the eSim method Virtual screening Pose prediction Multiple ligand alignment The core eSim methodology is also integrated into the Docking and QuanSA modules.

Features and benefits Virtual screening enrichment is both practically and statistically significantly better than alternative methods Virtual screening speeds of over 20 million compounds per day on a single computing core Databases of billions of molecules can be screened in hours using cloud-based computing resources Pose prediction accuracy is substantially better than alternative approaches

Docking and xGen Modules Top-Tier Solution for Virtual Screening and pose Prediction + Real-Space X-ray Density Modeling of Ligands

The Docking module addresses all aspects of ensemble docking Large-scale PDB retrieval and processing Surface-based binding site alignment using the PSIM method Fully automatic pocket variant selection to cover the relevant protein conformational variation Virtual screening Pose prediction

Feature and benefits Automated alignment and selection of appropriate binding site variants Robust and fully automatic modes for virtual screening and pose prediction\Very extensive validation Highly accurate non-cognate ligand docking Directly applicable to synthetic macrocycles, with accuracy equivalent to non-macrocycles The xGen module implements a novel method for real-space refinement and de novo fitting of ligand ensembles into X-ray density maps

Models ligand density using conformational ensembles Avoids atom-specific B-factors as X-ray model parameters Produces chemically sensible conformers with low strain energy; applicable to complex macrocycles Yields superior fit to X-ray density than standard fitting approaches Accessible to non-crystallographers and as part of crystallographic workflows

Affinity Module Unique Machine-Learning Approach for Prediction Binding Affinity and Pose

The Affinity Module implements the QuanSA (Quantitative Surface-field Analysis) method, which builds physically meaningful models that approximate the causal basis of protein ligand interactions. The module implements integrated procedures for quantitative prediction of both binding affinity and ligand pose, with or without protein structural information Multiple ligand alignment for molecular series that include multiple scaffolds Incorporation of known binding site information Machine-learning approach to physical binding site model induction using a multiple-instance approach Prediction of both binding affinity and binding mode of new ligands Iterative refinement of models with new data

Features and benefits Fully automatic model building, including all aspects of ligand conformation and alignment The binding site model (a “pocket-field”) is analogous to a protein binding site, including aspects of flexibility The pocket-field identifies which pose a new molecule must adopt, and ligand strain is directly modeled Measurements of prediction confidence and molecular novelty guide user interpretation Very detailed aspects of molecular surface shape, directional hydrogen bonding preferences, and Coulombic electrostatics are learned Requires as few as 20 molecules for model induction and is capable of modeling series of hundreds of molecules

What's New

本站不对文件进行储存，仅提供文件链接，请自行下载，本站不对文件内容负责，请自行判断文件是否安全，如发现文件有侵权行为，请联系管理员删除。