PSI Structural Biology Knowledgebase

PSI | Structural Biology Knowledgebase
Header Icons

Related Articles
Molecular Portraits of the Cell
February 2015
Transmembrane Spans
December 2014
Spot the pore
January 2010
How does Dali work?
December 2009
Learning from failure
December 2009
Face off
October 2009
Drug targets to aim for
February 2009

Technology Topics Dissemination Tools

How does Dali work?

PSI-SGKB [doi:10.1038/th_psisgkb.2009.55]
Technical Highlight - December 2009
Short description: Comparing three-dimensional protein structures can reveal functional clues. The Dali server is one of the most popular ways to achieve this.

DaliLite results page.

New structures are routinely scanned against those already in the Protein Data Bank (PDB) using one of a multitude of structural comparison servers in the hope that they will reveal similarities that will help to indicate a protein's function.

The traditional way to compare structures, and the method that many structural alignment programs use, is to treat each one as a rigid three-dimensional object and superimpose one on the other. Differences are calculated using a least-squares method.

The Dali server works differently, in that it uses a sum-of-pairs method, which produces a measure of similarity by comparing intramolecular distances. Similarity is measured by Dali-Z scores. Structures that have significant similarities have a Z-score above 2, and usually have similar folds.

DaliLite is a standalone software package, which users can download and run locally on their own computers, and it can also be used via the web to compare two structures.

With over 61,000 structures in the PDB, any search requires huge computing power. And so to speed up the process, the Dali server works by organizing structures ahead of time according to their fold. Originally, an internal database of processed PDB entries and their alignments was updated each week, until the sheer number of new structures began to overwhelm the system, leading to the PDB weekly update taking longer than a week to process.

To solve this problem, the update procedure was re-engineered so that similar structures are placed in a graph rather than recorded against every other structure, to speed up both the searching and the cataloguing of new structures.

The nodes of the graph represent protein structures and edges represent alignments. So instead of each structure being linked to all its nearest neighbors, it is now presented as a path of continuous structural similarity. This new mode is 30 times faster than the previous search mode.

But is DaliLite any good? In a recent example, six out of 32 programs or web servers that use a pairwise method of alignment correctly aligned the functional residues in urease and adenosine deaminase 1 . DaliLite was one of them. The others were SSAP, LGA/GDT, TOPOFIT, GASH and PPM.

Related articles

GPCR modeling: any good?

Protein modeling made easy

Model proteins in your lunch break

Maria Hodges


  1. H. Hasegawa and L. Holm. Advances and pitfalls of protein structural alignment.
    Curr. Opin. Cell Biol. 19, 341-348 (2009). doi:10.1016/

  2. L. Holm, S. Kääriäinen, P. Rosenström and A. Schenkel. Searching protein structure databases with DaliLite v.3. Bioinformatics.
    24, 2780-2781 (2008). doi:10.1093/bioinformatics/btn507

Structural Biology Knowledgebase ISSN: 1758-1338
Funded by a grant from the National Institute of General Medical Sciences of the National Institutes of Health