MISTIC2

MISTIC2 calculates and visualizes residue covariation within protein families to identify structurally or functionally constrained positions.

Key Features:

• Covariation methods: Implements Mutual Information per position (MIp), mean-field Direct Coupling Analysis (mfDCA), pseudo-likelihood maximization DCA (plmDCA), and Gaussian Direct Coupling Analysis (gaussianDCA).
• Method comparison and score combination: Enables comparison of outputs from different covariation methods and combination of scores for integrated analyses.
• Circos representation: Provides a circular layout summarizing alignment information and pairwise covariation.
• Covariation network visualization: Renders covariation data as a residue network to explore pairwise relationships.
• 3D structure mapping: Maps covariation scores onto protein 3D structures to relate covariation to spatial contacts.
• Sequence logo: Displays consensus sequences and residue variability within the protein family.
• Residue annotations: Associates detailed residue-level information with covariation metrics.
• ROC curve for contact prediction: Provides ROC curve analysis to evaluate accuracy of predicted residue contacts.
• Data tables and filtering: Offers tabular outputs and filtering capabilities for managing and selecting covariation data.
• Export capabilities: Allows exporting figures and result datasets for external analysis.
• Web service / API: Exposes programmatic access for automated covariation calculation and result retrieval.

Scientific Applications:

• Identification of correlated mutations: Detects correlated residue pairs across evolutionarily related proteins indicative of constraints on stability or function.
• Mapping functional and structural sites: Pinpoints positions crucial for protein structure or function by integrating covariation with structural data.
• Evaluation of contact prediction: Assesses residue contact prediction performance using ROC curve analyses.
• Analysis of evolutionary pressures: Characterizes evolutionary constraints and selective pressures acting on protein families.

Methodology:

Covariation is calculated using MIp, mfDCA, plmDCA, and gaussianDCA; scores can be compared and combined, and ROC curves are used to evaluate contact-prediction performance.