Machine Learning Center
Machine learning for the cryo-EM and cryo-ET pipeline
Simons Machine Learning Center
The Simons Machine Learning Center (SMLC) is a dedicated machine learning research group within NYSBC. The SMLC develops computational methods and software that improve how cryo-EM and cryo-ET data are collected, processed, and interpreted, with a focus on making structural biology workflows more automated, data-efficient, and informative.
SMLC research applies modern machine learning, computer vision, and deep learning approaches to challenges across the cryo-EM pipeline. Areas of emphasis include automated microscope targeting and data collection; particle picking and object detection; image denoising; analysis of heterogeneous and flexible structures; tomogram and cellular segmentation; and models that integrate information from cryo-EM, cryo-ET, protein sequence, structure, and function. These methods help researchers extract more biological insight from complex, noisy, and information-rich datasets.
By pairing algorithm development with the experimental scale of SEMC, the SMLC creates tools that can be tested on real cryo-EM and cryo-ET workflows and then shared with the broader structural biology community. This close connection between computation, instrumentation, and biological application allows the group to address practical bottlenecks in data acquisition and analysis while advancing new approaches for understanding biomolecular structure and function.
Publications
Ghostbuster: Beyond Ewald Sphere Curvature Correction for Cryo-EM Single Particle Reconstruction Via Phase Retrieval
Ghostbuster: Beyond Ewald Sphere Curvature Correction for Cryo-EM Single Particle Reconstruction Via Phase Retrieval
Latent disentanglement in mesh variational autoencoders improves the diagnosis of craniofacial syndromes and aids surgical planning
Computer Methods and Programs in Biomedicine, Vol. 256, Pages 108395
Ghostbuster: A phase retrieval diffraction tomography algorithm for cryo-EM
Ultramicroscopy, Vol. 262, Pages 113962
The value of Augmented Reality in surgery — A usability study on laparoscopic liver surgery
Medical Image Analysis, Vol. 90, Pages 102943
Machine learning optimization of candidate antibody yields highly diverse sub-nanomolar affinity antibody libraries
Nature Communications, Vol. 14, Issue 1
Learning to automate cryo-electron microscopy data collection with Ptolemy
IUCrJ, Vol. 10, Issue 1, Pages 90-102
Machine Learning Optimization of Candidate Antibodies Yields Highly Diverse Sub-nanomolar Affinity Antibody Libraries
Automatic, global registration in laparoscopic liver surgery
International Journal of Computer Assisted Radiology and Surgery, Vol. 17, Issue 1, Pages 167-176
An engineered protein-phosphorylation toggle network with implications for endogenous network discovery
Science, Vol. 373, Issue 6550
CryoDRGN: reconstruction of heterogeneous cryo-EM structures using neural networks
Nature Methods, Vol. 18, Issue 2, Pages 176-185
DeepSeqPanII: an interpretable recurrent neural network model with attention mechanism for peptide-HLA class II binding prediction
IEEE/ACM Transactions on Computational Biology and Bioinformatics, Vol. 19, Issue 4, Pages 2188-2196
Deep learning pan-specific model for interpretable MHC-I peptide binding prediction with improved attention mechanism
Proteins: Structure, Function and Bioinformatics, Vol. 89, Issue 7, Pages 866-883
Machine learning across the cryo-EM pipeline—targeting particle-picking, denoising, segmentation.