SBGrid Consortium homepage

cryoCAT is an open source, GPL-3.0+ licensed library providing contextual analysis tools for cryo electron tomography and subtomogram averaging.

a pipeline of tools to generate robust open mesh surfaces from voxel segmentations of biological membranes using the Screened Poisson algorithm and calculate morphological features including curvature and membrane-membrane distance using pycurv's vector voting framework with tools to convert these morphological quantities into morphometric insights.

a new web-based user interface for CCP-EM and RELION software.

a supervised machine learning method to directly reconstruct the tomogram from aligned cryo-ET tilt series. The methods is trained on real measurements using FBP+cryo-CARE+IsoNet reconstructions as the reference.

an integrated suite of software tools for processing single particle cryoEM data, from preprocessing raw image data through building and fitting atomic models.

a collection of cheminformatics and machine-learning software written in C++ and Python.

a collection of scripts for cryoEM data collection and analysis

is a biomolecular structure prediction model aiming to be a bitwise reproduction of DeepMind's AlphaFold3, developed by the AlQuraishi Lab at Columbia University and the OpenFold consortium. It supports structure prediction of standard and non-canonical protein, RNA, and DNA chains, and small molecules, with pipelines for generating MSAs using ColabFold server or JackHMMER/hhblits following the AlphaFold3 protocol. The tool includes kernel acceleration through cuEquivariance and DeepSpeed4Science kernels, multi-query job support with distributed predictions across multiple GPUs, and performs competitively with state-of-the-art open source biomolecular structure prediction models.

is an atom-level enzyme active site scaffolding tool that generates protein backbones conditioned on atomic motifs and small molecules. The software supports enzyme design from atomic motifs with unknown sequence positions, small-molecule binder design with RASA conditioning, and includes integration with LigandMPNN for sequence fitting and Chai1 for structure validation. It provides both indexed backbone and unindexed atomized region outputs with optional idealization steps for enhanced structural quality.

is a deep learning-based tool for designing protein and peptide binders against target proteins or small molecules. The software uses diffusion models to generate diverse binder candidates, followed by inverse folding, structure prediction with Boltz-2, and comprehensive filtering to select high-quality designs. It supports multiple protocols including protein-anything, peptide-anything, nanobody-anything, and protein-small_molecule design, with automated pipelines that can generate thousands of candidates and rank them based on binding affinity and structural quality metrics.