Mark Basham

Marjan Famili, Jola Mirecka, Camila Rangel Smith et al.

Learning compact and interpretable representations of data is a critical challenge in scientific image analysis. Here, we introduce Affinity-VAE, a generative model that enables us to impose our scientific intuition about the similarity of instances in the dataset on the learned representation during training. We demonstrate the utility of the approach in the scientific domain of cryo-electron tomography (cryo-ET) where a significant current challenge is to identify similar molecules within a noisy and low contrast tomographic image volume. This task is distinct from classification in that, at inference time, it is unknown whether an instance is part of the training set or not. We trained affinity-VAE using prior knowledge of protein structure to inform the latent space. Our model is able to create rotationally-invariant, morphologically homogeneous clusters in the latent representation, with improved cluster separation compared to other approaches. It achieves competitive performance on protein classification with the added benefit of disentangling object pose, structural similarity and an interpretable latent representation. In the context of cryo-ET data, affinity-VAE captures the orientation of identified proteins in 3D which can be used as a prior for subsequent scientific experiments. Extracting physical principles from a trained network is of significant importance in scientific imaging where a ground truth training set is not always feasible.

Beatriz Costa-Gomes, Joel Greer, Nikolai Juraschko et al.

Ease of access to data, tools and models expedites scientific research. In structural biology there are now numerous open repositories of experimental and simulated datasets. Being able to easily access and utilise these is crucial for allowing researchers to make optimal use of their research effort. The tools presented here are useful for collating existing public cryoEM datasets and/or creating new synthetic cryoEM datasets to aid the development of novel data processing and interpretation algorithms. In recent years, structural biology has seen the development of a multitude of machine-learning based algorithms for aiding numerous steps in the processing and reconstruction of experimental datasets and the use of these approaches has become widespread. Developing such techniques in structural biology requires access to large datasets which can be cumbersome to curate and unwieldy to make use of. In this paper we present a suite of Python software packages which we collectively refer to as PERC (profet, EMPIARreader and CAKED). These are designed to reduce the burden which data curation places upon structural biology research. The protein structure fetcher (profet) package allows users to conveniently download and cleave sequences or structures from the Protein Data Bank or Alphafold databases. EMPIARreader allows lazy loading of Electron Microscopy Public Image Archive datasets in a machine-learning compatible structure. The Class Aggregator for Key Electron-microscopy Data (CAKED) package is designed to seamlessly facilitate the training of machine learning models on electron microscopy data, including electron-cryo-microscopy-specific data augmentation and labelling. These packages may be utilised independently or as building blocks in workflows. All are available in open source repositories and designed to be easily extensible to facilitate more advanced workflows if required.

Mark Basham

Diamond Light Source is the UK's National Synchrotron Facility and as such provides access to world class experimental services for UK and international researchers. As a user facility, that is one that focuses on providing a good user experience to our varied visitors, Diamond invests heavily in software infrastructure and staff. Over 100 members of the 600 strong workforce consider software development as a significant tool to help them achieve their primary role. These staff work on a diverse number of different software packages, providing support for installation and configuration, maintenance and bug fixing, as well as additional research and development of software when required. This talk focuses on one of the software projects undertaken to unify and improve the user experience of several experiments. The "mapping project" is a large 2 year, multi group project targeting the collection and processing experiments which involve scanning an X-ray beam over a sample and building up an image of that sample, similar to the way that google maps bring together small pieces of information to produce a full map of the world. The project itself is divided into several work packages, ranging from teams of one to 5 or 6 in size, with varying levels of time commitment to the project. This paper aims to explore one of these work packages as a case study, highlighting the experiences of the project team, the methodologies employed, their outcomes, and the lessons learnt from the experience.

Nicola Wadeson, Mark Basham

Diamond Light Source (DLS), the UK synchrotron facility, attracts scientists from across the world to perform ground-breaking x-ray experiments. With over 3000 scientific users per year, vast amounts of data are collected across the experimental beamlines, with the highest volume of data collected during tomographic imaging experiments. A growing interest in tomography as an imaging technique, has led to an expansion in the range of experiments performed, in addition to a growth in the size of the data per experiment. Savu is a portable, flexible, scientific processing pipeline capable of processing multiple, n-dimensional datasets in serial on a PC, or in parallel across a cluster. Developed at DLS, and successfully deployed across the beamlines, it uses a modular plugin format to enable experiment-specific processing and utilises parallel HDF5 to remove RAM restrictions. The Savu design, described throughout this paper, focuses on easy integration of existing and new functionality, flexibility and ease of use for users and developers alike.

Imanol Luengo, Mark Basham, Andrew P. French

Recent applications in computer vision have come to heavily rely on superpixel over-segmentation as a pre-processing step for higher level vision tasks, such as object recognition, image labelling or image segmentation. Here we present a new superpixel algorithm called Hierarchical Piecewise-Constant Super-regions (HPCS), which not only obtains superpixels comparable to the state-of-the-art, but can also be applied hierarchically to form what we call n-th order super-regions. In essence, a Markov Random Field (MRF)-based anisotropic denoising formulation over the quantized feature space is adopted to form piecewise-constant image regions, which are then combined with a graph-based split & merge post-processing step to form superpixels. The graph and quantized feature based formulation of the problem allows us to generalize it hierarchically to preserve boundary adherence with fewer superpixels. Experimental results show that, despite the simplicity of our framework, it is able to provide high quality superpixels, and to hierarchically apply them to form layers of over-segmentation, each with a decreasing number of superpixels, while maintaining the same desired properties (such as adherence to strong image edges). The algorithm is also memory efficient and has a low computational cost.