BioMedBridges: Final version of the web-service for segmentation and identification of components in volume data

We have developed a software pipeline, named SMaSB, to perform automated volume/shape matching. Together with a database of over 1000 volumes from the Electron Microscopy Data Bank (EMDB) and Protein Data Bank (PDB), the pipeline underpins a web service PDBeShape. These tools are novel in providing access to a growing class of structural biology data viz. volume data. The development of SMaSB was reported in Deliverable 9.1, and the development of PDBeShape in Deliverable 9.2.

Deliverable 9.3 implements automated and manual segmentation into the PDBeShape web service. Automated segmentation with Chimera-Segger is now an integral part of volume pre-processing, with the results output as Chimera-Segger .seg files. Automated segmentation is useful as a guide to features in a volume, but does not reliably give segments corresponding to biological components (such as protein or RNA chains). Therefore, PDBeShape also supports annotation with the results of manual segmentation. As a proof of principle, 86 entries from the PDBeShape volume database were manually segmented and annotated, and the results are available from the PDBeShape service. Biological annotation links volume segments to entries in Pfam / Rfam and Uniprot.

Segmentation of structural volume data is an area of intense discussion in the international community. Issues include the reliability of algorithms, the representation and annotation of results, and the practicalities of sharing the results. The EBI hosted a workshop on “3D segmentations and transformations - building bridges between cellular and molecular structural biology” on 7^th / 8^th December 2015. Work from the current deliverable was presented as an example user case, and will lead to further work in this area.