CATHedral suite



Structure based sequence alignment using double dynamic programming.


Rapid database scanning.


Pairwise and multiple alignments.

ebisu group

chemogenomix © 2012

Read more below

CATHedral Scan

When you have a either a new structure or an existing one that you wish to take forward as a drug discovery target it is sensible to first search the data base of know structures to find relevant similarities.

With a version of the PDB stored locally you can first implement the scanning mode. This initially looks for more general similarities ensuring that more detailed pairwise calculations (see below) are only spent on comparisons with a reasonable prospect of having meaningful similarity.


CATHedral Pairwise

CATHedral delivers high quality pairwise alignments due to the inclusion of side-chain vector information in the calculation. This ensures that buried residues do not get aligned with accessible ones.

Research Notes

Automated pairwise structure alignment can be implemented by many different algorithms, but the first was based on double dynamic programming by Orengo and Taylor. This approach underpins the CATHedral pairwise program.

For more information download this publication:

Case Study

Three distantly related proteins osmolarity sensor protein (1byq) DNA gyrase (1z5a) and HSP90 (1byq) only have ~10% residue identity over their entire domains nucleotide binding domain, yet their ATP binding sites are highly conserved and the regions of high conservation are clearly clustered.

All three proteins have been considered as potential drug targets; DNA gyrase is the target for novobiocin, HSP90 compounds are in clinical trials and the osmolarity sensor protein was considered as an antimicrobial target. Such observations have key implications to both off-target effects and compound reuse.

See Case Studies section for detailed information

Availability

Linux

CATHedral Multi

Once you have found all of your pairwise hits, you will probably want to generate a multiple, structure-based sequence alignment. Even in sequence space, multiple alignments are difficult to achieve so the algorithmic challenge is inevitable more complicated for multiple structure alignment.

As a consequence automated multiple structure alignment methods are few but fortunately the CATHedral suite has one and is included in the suite of programs. It is important to ensure that your entries have the same number of domains. In fact we typically recommend concentrating on the target domain of interest unless you are particularly interested in multiple domains.

The example above shows the structure based sequence alignment for the same example shown on the Ligplot+ page where very distant homologues are aligned.