Prof. Dr. Maya Topf got her PhD degree at the University of Oxford in 2002 where she used hybrid QM/MM and classical molecular dynamics simulations to study serine protease deacetylation. She then moved to the University of San Francisco to join Andrej Sali’s lab (don’t miss their last paper on IMP also highlighted in this newsletter!) where she successfully combined cryo-electron microscopy data and comparative modelling to model macromolecular structures at an atomic resolution.

Later on, Maya came back to UK at the Institute of Structural and Molecular Biology and Birkbeck, where she could further work at the interface between bioinformatics and cryo-electron microscopy. Over the last years, the Topf group developed numerous methods to integrate experimental data into the modelling process of macromolecular assemblies, focusing on 3D-EM and mass spectrometry data. Among them, they released TEMPy, a python library to assess the fits of atomic models in 3D electro-microscopy density maps …

1) Can you introduce your software in a few words?

TEMPy is a python package for integrative modelling primarily based on cryo electron microscopy density (but also enables integration of data from cross-linking coupled to mass-spectrometry) (J App Cryst 2015, 48:1314, Mol Cell Proteomics 2016, 15:10.1074). Simultaneous fitting of components of an assembly is carried out by a genetic algorithm, which uses a mutual information metric to evaluate density fit and penalises for steric clashes (Structure 2016, 23:2365). TEMPy integrates various scoring functions, including correlation-based scores, surface based-scores and statistical scores (J Struct Biol, 2017, 199:12–26), to assess the models built by integration of multiple data.

2) Regarding the development model of your software, who is in charge of the development/maintenance/support?

Many people in the Topf group have contributed to the software. At the moment, the main developer and support person is Dr. Agnel Praveen Joseph.

3) In which regards does your software fall within the field of integrative modelling?

TEMPy has been primarily designed to work on cryo-electron microscopy data and for building and evaluating atomic models based on this data. This is by definition an integrative approach. We have developed TEMPy in a very modular way, to deal with diverse sources of information (e.g., EM density and distance restraints based on cross-linking data). We are working on ways of including other experimental data, and choosing an adequate representation and formulation of the restraints to represent the data as accurately as possible. 

4) Can you share with us an example in which the use of your software was key to answer a scientific question?

In a recent application, low-resolution density from sub-tomogram averaging was combined with chemical cross-linking to identify the oligomerization state of the HSV-1 glycoprotein B assembly in vesicle membranes. Multiple methods of hierarchical constrained density-fitting were employed for building atomic models for the pre- and post-fusion conformations (Proc Natl Acad Sci U S A 2016, 113:4176). Models were generated and scored by TEMPy in the context of an ensemble of alternate conformations and validated using mutation data, revealing a surprising pre-fusion conformation. In another example we have used it to help us understand conformational changes involved in pore formation by the perforin-related Fungal toxin pleurotolysin, by generating and scoring multiple models representing intermediate states observed by cryo electron microscopy (PLoS Biol 2015, 13(2): e1002049).