G-LoSA (Graph-based Local Structure Alignment) is an efficient computational method to align protein local structures in a sequence order independent way and to provide the GA-score (between 0 and 1), a size-independent quantity of structural similarity for a given local structure pair. In particular, the GA-score is calculated based on the chemical features of each amino acid and can be applied to measure the structural similarity with the local structures of diverse sizes and characteristics, yet maintaining its length independency. In G-LoSA, both the iterative maximum clique search and the fragment superposition are adopted to generate all possible alignments between two different local structures and the optimal alignment is determined by the maximum GA-score (figure below).

1. Installation

Click here to download the G-LoSA software package including example PDB local structures. The source code of G-LoSA (glosa.cpp) is written in C++ and can be easily compiled using C++ GNU compiler by
>g++ -c glosa.cpp
>g++ -o glosa glosa.o

2. Preparing Chemical Feature and Secondary Structure Files

Input protein structure files for G-LoSA must be in PDB format and be ended with "TER".

G-LoSA needs chemical feature file for each input protein structure to calculate GA-score. Users can also optionally use a secondary structure file for each input protein structure to reduce computational time. The G-LoSA software package contains auxiliary programs (AssignChemicalFeatures.java and AssignSecondaryStructures.java) to easily generate these chemical feature and secondary structure files.

* For chemical feature information
>javac AssignChemicalFeatures.java (for compile)
>java AssignChemicalFeatures [structure file in PDB format] (for execution)

* For secondary structure information
>javac AssignSecondaryStructures.java (for compile)
>java AssignSecondaryStructures [local structure file in PDB format] [global structure file in PDB format] (for execution)

3. Running G-LoSA

-s1 structure 1 (PDB format)
-s1cf chemical feature file for structure 1
-s2 structure 2 (PDB format)
-s2cf chemical feature file for structure 2
[advanced options]

4. Outputs


G-LoSA alignment score


PDB coordinates of s2 structure aligned onto s1 structure


translational and rotational matrix to align -s2 structure onto -s1 structure


PDB coordinates of s2w structure transferred by the matrix


G-LoSA Toolkit

Please visit here for G-LoSA Toolkit.


  • Hui Sun Lee and Wonpil Im, G-LoSA: An Efficient Computational Tool for Local Structure-Centric Biological Studies and Drug Design. Protein Sci. 25:865-876 (2016).
  • Hui Sun Lee and Wonpil Im, Ligand Binding Site Detection by Local Structure Alignment and Its Performance Complementarity. J. Chem. Inf. Model. 53:2462-2470 (2013).
  • Hui Sun Lee and Wonpil Im, Identification of Ligand Templates using Local Structure Alignment for Structure-based Drug Design. J. Chem. Inf. Model. 52:2784-2795 (2012).


  • G-LoSA and G-LoSA Toolkit are freely available to all academic users and not-for-profit institutions.
  • For commercial users, please contact us (wonpil@ku.edu or huisun@ku.edu) to get the license of using this distribution.

G-LoSA Update History

* For any question or suggestion, e-mail to Hui Sun Lee (huisun.cadd@gmail.com)