IDSS: deformation invariant shape signatures of flexible proteins

Yu-Shen Liu1    Yi Fang1    Karthik Ramani1,2    

1School of Mechanical Engineering
2School of Electrical Computer Engineering (by courtesy)
PurdueUniversity, West Lafayette, IN, 47907, USA

 

 

 

Abstract

Comparison of proteins shapes is receiving more attention for understanding evolutionary relationships and for the prediction of function. Many existing shape-based protein comparison methods are effective for comparing rigid objects, but they can not handle shape deformation of flexible objects well. Nevertheless, proteins are flexible molecules that undergo significant structural changes and shape deformation as part of their function.

We introduce the Inner Distance Shape Signature (IDSS) for describing the 3D shape of flexible proteins. The inner distance is defined as the length of the shortest path between landmark points within the shape, and it reflects well the protein structure and deformation without explicitly decomposition. Our IDSS is stored as a histogram which is a probability distribution of inner distance between all sample point pairs. We show that IDSS is insensitive to shape deformation of flexible proteins and more effective at capturing protein structures than traditional shape signatures. Our approach reduces the 3D shape comparison problem of flexible proteins to the comparison of IDSS. We evaluate and demonstrate the effectiveness of IDSS within a protein search engine application for a benchmark containing abundant conformational changes.

Downloads
  1. README.pdf
  2. Tutorial.pdf
  3. IDSS_win32exe.zip
  4. Database (~7.5MB) 
  5. Demo.avi (~29MB) ,  Demo.swf (~29MB)

Examples

 

      

Our inner distance (ID) signature is compared, for instance, to Euclidean distance (ED) signature from [Osada02]. The first row shows the input four artificial proteins with the same main chain orientation but with different surface shapes. The second row shows the ID and ED signatures. In each plot, the vertical axis represents distance distribution. Note that ID is not sensitive to shape deformation, so four signatures are almost consistent; in contrast, ED is strongly sensitive to deformation.

 

 

   

A screen shot from flexible proteins shape retrieval system using our IDSS algortihm. In this figure, the query protein's information including its name and Group ID is shown on the top-left, and the approximated shape based on inner distance is displayed. The query example from Group 1 has four deformation shapes, and the search system returns the first four proteins from the same group

 

Acknowledgment

The database of Macromolecular Movements (MolMovDB) used in this paper is provided by Mark B Gerstein and Samuel C Flores. This material is partly based upon work supported by the National Science Foundation under Grant IIS No. 0535156. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation. We also acknowledge partial support from the National Institute of Health (GM-075004).