MULTICOM: An Accurate Multi-Template Comparative Modeling Server for Protein Structure Prediction

Introduction

MULTICOM is an accurate protein structure prediction server using comparative (or homology) modeling with multi-template combination. The goal is to generate high-quality models for protein targets having homologous template proteins in Protein Data Bank (PDB). The template library is updated daily according to the new release of PDB. MULTICOM makes predictions in three steps.

Step 1: Use PSI-BLAST to search the target sequence against Non-Redundant protein sequence database to generate a profile. Then search the profile against the template library to identify templates. If some templates are found, MUTCOM will generate a 3D structure model as in the following two steps. Otherwise, it stops and no model is generated. In the latter case where the target is most likely not suitable for comparative modeling, user is advised to use fold recognition (or threading) servers.

Step 2: If more than one templates are found, significant templates are combined. The pairwise alignments between the target and templates produced by PSI-BLAST are combined into a multiple alignment.

Step 3: Based on the target-template alignments and 3D structures of the templates, Modeller (Sali and Blundell, 1993) is used to generate a structure model for the target protein. The model in PDB format and the alignment in PIR format are attached with the returned email. The PDB code, e-evalue, and identity of templates are also included.

Input and Output

Inputs to the web server include target name, sequence, and email address. It usually takes several minutes to process one query, depending on the server load and sequence length. Sequence must be entered as a plain sequence of amino acids. Maximum sequence length allowed is 1500.

Normal outputs include a 3D model in PDB format, the corresponding alignment file in PIR format, and the list of templates with e-value and sequence identity. If no templates is found, no model will be generated. The target may be not suitable for comparative modeling. User is advised to use fold recognition (or threading) servers.