ala-tay / aitom

AI for tomography

AITom

Introduction

AITom is an open-source platform for AI driven cellular electron cryo-tomography analysis.

AITom is originated from the tomominer library, adapted from an extended version of the tomominer library, developed at Alber Lab.

Tutorials

Install

About us

Xulab at Carnegie Mellon University Computational Biology Department

Code and data for projects developed and maintained by Xu Lab and collaborators.

The research related to the code and data can be found at http://cs.cmu.edu/~mxu1

Background

Nearly every major process in a cell is orchestrated by the interplay of macromolecular assemblies, which often coordinate their actions as functional modules in biochemical pathways. To proceed efficiently, this interplay between different macromolecular machines often requires a distinctly nonrandom spatial organization in the cell. With the recent revolutions in cellular Cryo-Electron Tomography (Cryo-ET) imaging technologies, it is now possible to generate 3D reconstructions of cells in hydrated, close to native states at submolecular resolution.

Research

We are developing computational analysis techniques for processing large amounts of Cryo-ET data to reconstruct, detect, classify, recover, and spatially model different cellular components. We utilize state-of-the-art machine learning (including deep learning) approaches to design Cryo-ET specific data analysis and modeling algorithms. Our research automates the cellular structure discovery and will lead to new insights into the basic molecular biology and medical applications.

De novo structural mining pipeline results: (a). A slice of a rat neuron tomogram, (b). Recovered patterns (from left to right): mitochondrial membrane, Ribosome-like pattern, ellipsoid of strong signals, TRiC-like pattern, borders of ice crystal, (c). Pattern mining results embedded, (d). Individual patterns embedded.

Cite AITom

Funding