About the Volume Browser
Three dimensional electron microscopy (3DEM) produces intricate structural data permitting visualisation of multiple complementary perspectives. The Volume Browser is an online tool that harnesses these perspectives into a singular view to offer a better understanding of visual data.
The Volume Browser (VB) is an online tool for visualising 3DEM structures by bringing together several visual data sets: EM images, segmentations (in 2D and 3D), 3D isosurfaces and molecular models (from the PDB). Additionally, the VB displays structured annotations from various ontologies while also providing links to related archives. The goal of the VB is to enrich datasets by creating relevant links between entries thereby facilitating exploration of entries.
Anatomy of the Volume Browser
|1||Entry Information||Provides a brief description of the underlying entry. The entry information comprises a link to the actual entry, the title of the entry, details about the experimental modality (e.g. SXT, FIB-SEM) and details such as resolution and species.|
|2||Visualisation Panel||In addition to slices through the 3D volume, the visualisation panel provides four perspectives: top (viewed against the Z axis), front (viewed against the X axis) and right side (viewed against the Y axis) and a 3D view. We overlay contours representing slices through the segmentation in each of the first three views, which the user may choose to hide. These segmentation overlays are interactive: users may click each segment overlay to view related biological entities presented in the Segments Table (see 3 below).|
|3||Segments Table||Each segment is annotated with accessions to publicly accessible biological terms or entities. This component lets users interact with the overlaid segments by displaying the biological context of the current segment. The last panel (Download segmentation) provides links to download the actual segmentations.|
|4||Slice Navigator||A simple and intuitive widget for users to navigate through the volume’s slices. There is one slider per axis and the current slice and zoom are properly oriented.|
|5||3D Viewer||The 3D Viewer currently displays an isosurface for non-tomographic entries. It uses the recommended contour level but different levels can be selected as well. We will develop this to also display the 3D segmentation.|
|6||Zoom||Zooming can be step-wise (using buttons) or direct (using the slider).|
|7||Opacity||Allows varying segment transparency. The segment outlines are retained for context.|
Why does the Volume Browser exist anyway?
The Volume Browser is the solution to two problems: the entry silo problem and the untraceable segmentation problem. We will explain each in turn.
The Entry Silo Problem
Every week close to 100 new entries are deposited into the both EMDB and EMPIAR and this number continues to grow year on year. Unfortunately, the only relationships between entries are similarity by deposition despite the fact that one can easily find hundreds of entries related by biologically relevant categories (e.g. species, tissue, homology etc.) Each entry or set of entries related by deposition exist independent of other entries - in a silo.
And this is not confined to the two archives. There are related archives that host relevant data such as protein accessions (such as UniProt, Protein Resource), functional annotations (such as the Gene Ontology), molecular models (such as the Protein Data Bank), raw data (EMPIAR), and literature (Europe PMC). The number of such links can only grow with time.
The Volume Browser solves this problem by exposing segmentation annotations made using accessions from these resources admitting richer insight into related biological categories.
The Untraceable Segmentation Problem
Segmentations are expensive derivatives of images because they are often tedious to generate. Presently, segmentations only appear in publications to visually highlight biological semantics in structures. Thereafter they are very difficult to track down usually because the person who created them (usually a student or postdoc) has moved on!
This problem is solved by having the Volume Browser act as a repository for segmentations that may be consumed alongside the image data. The segmentations are stored using the open EMDB-SFF which will eventually be convertible to common application-specific file formats.
Where does the data come from?
Each VB entry consists of two key pieces of data:
- the image data is derived from a 3D density map of an entry deposited in either EMDB or EMPIAR, and
- an annotated segmentation saved as an EMDB-SFF file (XML/HDF5)
The annotated segmentation must spatially correspond to the image data i.e. is the segmentation of the image data otherwise it will result in poor or no registration to the image.
Is the data available for download?
Yes. You can download the annotated segmentation from under the Download section of the segment table.
What can I do with the data?
Presently, the only tool that works directly with EMDB-SFF files is the EMDB-SFF toolkit (sfftk). sfftk is a Python command-line tool and API to work with EMDB-SFF files. Our hope is that sfftk will then be incorporated into other segmentation tools so that eventually you will be able to view segmentations in your favourite segmentation tool.
How can I create a VB entry?
As mentioned in Where does the data come from? each VB entry has both image and annotated segmentation. If you have already deposited the entry into EMDB or EMPIAR then all you will need to add is an annotated EMDB-SFF segmentation. There are two ways to create an annotated EMDB-SFF:
- using sfftk, or
- using the Segmentation Annotation Tool (SAT).
Each method provides detailed descriptions on how to create annotated EMDB-SFF segmentations.