Multiple Mapping Method with Multiple Templates (M4T): optimizing sequence-to-structure alignments and combining unique information from multiple templates András Fiser Department of Biochemistry and Seaver Center for Bioinformatics Albert Einstein College of Medicine Bronx, New York, USA
23
Embed
Andr á s Fiser Department of Biochemistry and Seaver Center for Bioinformatics
Multiple Mapping Method with Multiple Templates (M4T): optimizing sequence-to-structure alignments and combining unique information from multiple templates. Andr á s Fiser Department of Biochemistry and Seaver Center for Bioinformatics Albert Einstein College of Medicine Bronx, New York, USA. - PowerPoint PPT Presentation
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Multiple Mapping Method with Multiple Templates (M4T): optimizing sequence-to-structure alignments and combining
unique information from multiple templates
András Fiser
Department of Biochemistry andSeaver Center for BioinformaticsAlbert Einstein College of MedicineBronx, New York, USA
Target – TemplateAlignment
Model Building
START
Template Search
Model Evaluation
END
Multiple Mapping Method
Loop, side chain modeling
Statistical potential
Comparative protein structure modeling
Multiple Templates
Why do we need sequence alignments?
#Sequence vs. structure To generate input alignment for comparative modeling / threading
#Sequence vs. databases:
Querying sequence databases
#Sequence vs. sequence:
Establishing residue equivalencies between two proteins to locate conserved/variable regions
Ranking of models built on alternative alignments
Problem: None of the currently available methods produce consistently superior results in all cases
2. A scoring scheme based on a comparison (PHD vs. DSSP) of the secondary structure types (H3P22).
3. Statistically derived residue-residue contact energy (Rykunov and Fiser3).1Shi et al. J. Mol. Biol. (2001) 310, 243-2572Rice et al., J. Mol. Biol (1997) 267, 1026-10383Rykunov & Fiser., Proteins. (2007) 67, 559-68
MMM performance on 1400 pairs
MMM performance on 87 pairs, meta-servers
ESypred3D Consensus
Sampling vs. Scoring
• Multiple Mapping Method optimally combines alternative alignments obtained from different methods or scoring function:
On a benchmark dataset of 6635 protein pair structural alignments, comparative models built using MMM alignments are approximately 0.3 Ǻ and 0.5 Å more accurate on average in the whole spectrum and in the <30% target-template sequence identity regions, respectively, than the average accuracy of models built using the alternative input alignments ( ~3 and ~4 Å).
Summary
Optimally combining multiple templates
Selecting multiple templates
• Target sequence: by PSI-BLAST.
• Hits selected if sequence overlap with the target is > 60% of the actual SCOP domain length or more than 75% of the PDB chain length in case of a missing SCOP classification.
• Iterative clustering procedure identifies the most suitable templates to combine. Templates are selected or discarded according to a hierarchical selection procedure that accounts for – sequence identity between templates and target sequence,
– sequence identity among templates,
– crystal resolution of the templates,
– contribution of templates to the target sequence (i.e. if a region is covered by several templates or by a single template only).
Single versus multiple templatesUsing a dataset of 765 proteins with known structure two sets of models were built: (1) using one template (best E-value hit; light bars), (2) using multiple templates (grey bars)
And…increased coverageHistogram of models’ difference length. Each query sequence is modeled using single and multiple templates. The histogram shows the frequency of (Lm–Ls). Lm: length of model built using multiple templates, and Ls length of the model built using a single template.
The x-ray structure, the model with multiple templates and with a single template are shown in grey, red, and blue, respectively.
Multiple templates agree much better in two exposed regions: A and B, than the model built using single template.
Increased Coverage
The x-ray structure, the model with multiple templates, and model with single templates are shown in grey, red, and blue, respectively.
The addition of extra templates allowed obtaining a longer model that include a beta-turn-beta-turn extra region (20 amino acids), depicted in ribbon.