Mohona: Where sequence, structure and pathway meet

মোহনা · confluence, where rivers meet

Mohona retrieves protein structures from the Protein Data Bank and the AlphaFold Protein Structure Database, relates each residue to its codon, assembles complexes and characterises amino acid substitutions.

Try

Start with an example

Capabilities

Sequence and structureThe sequence is linked to the structure, down to each codon of the gene. Oligomer assemblyOligomers are assembled on templates from the Protein Data Bank. Prediction of complexesInput files are prepared for AlphaFold 3, AlphaFold Server, Boltz-2 and OpenFold3. Pathways and interaction partnersReactions, pathway steps, functional sites and interaction partners are reported. Metabolon modellingEnzymes are selected, co-folded and analysed for interfaces and inter-site distances. Mutational analysisSubstitutions are characterised, with ΔΔG predicted by DDMut.

Data come straight from RCSB PDB, PDBe, AlphaFold DB, UniProt, ENA, STRING, Foldseek and PubChem. About Mohona

About

Mohona মোহনা

Mohona is a web application for the visualisation and analysis of protein structures and sequences. It integrates experimental structures, predicted models, sequence annotation, metabolic pathways and structure prediction in a single workspace.

Capabilities

Mohona displays entries from the Protein Data Bank and models from the AlphaFold Protein Structure Database, with the amino acid sequence linked to the three-dimensional structure.
It presents the coding sequence of each protein, with every residue aligned to its codon.
It superposes predicted models on experimental structures and reports the root-mean-square deviation.
It assembles homo-oligomers and hetero-oligomers on templates from the Protein Data Bank, using proteins from the same organism.
It prepares input files for the prediction of complexes with AlphaFold 3, AlphaFold Server, Boltz-2 and OpenFold3.
It reports catalytic reactions, pathway steps, functional sites and interaction partners for each protein.
It supports the modelling of metabolons, from the selection of enzymes to the analysis of interfaces and inter-site distances.
It characterises amino acid substitutions and reports predicted changes in stability and binding affinity.

Name and logo

Mohona (মোহনা) is the Bangla word for a confluence, the point at which rivers meet. The logo represents three subunits that assemble into a complex, as streams converge in an estuary.

Intended use

Mohona is intended for research in structural biology, enzymology and drug discovery. It supports the examination of individual proteins, the modelling of complexes and the structural interpretation of sequence variation.

Data resources

RCSB PDB and PDBe: Experimental structures, biological assemblies, ligands and entry annotations are retrieved from these archives.
AlphaFold Protein Structure Database: Predicted models are retrieved with their pLDDT values and predicted aligned error.
UniProt: Protein names, functions, functional sites, catalytic reactions, pathways and sequence variants are retrieved from UniProt.
European Nucleotide Archive: The coding sequence of each protein is retrieved from the European Nucleotide Archive.
STRING: Interaction partners are retrieved with the evidence that supports each association.
Foldseek: Structural similarity searches identify templates for oligomer assembly.
OMA, OrthoDB and eggNOG: Orthology groups identify partner subunits in the same organism.
PubChem: Compound names and identifiers support the ligand search.
DDMut and DDMut-PPI: These predictors estimate the effect of substitutions on folding stability and on binding affinity.
KEGG, BioCyc, Reactome, Rhea, BRENDA, InterPro and Complex Portal: Each protein is linked to its records in these resources.

Data handling

Public data are retrieved directly from the source databases, and search terms are sent to these databases as they are typed. Structure files opened from a computer are read locally in the browser.

Two functions transmit a structure to an external service at the request of the user. The template search sends the AlphaFold model to the Foldseek server, and the stability analysis sends coordinates to the DDMut servers at the University of Queensland.

Structure prediction

Mohona writes input files for AlphaFold 3, AlphaFold Server, Boltz-2 and OpenFold3, and provides the command for each predictor. Predictions run on AlphaFold Server or on a computer on which the predictor is installed.

The predicted model and its confidence file are then opened in Mohona for analysis.

Methods

Sequence and structure mapping: Chains of PDB entries are mapped to UniProt sequences with the SIFTS alignment. AlphaFold models follow UniProt numbering.
Coding sequence: The coding sequence cross-referenced by UniProt is translated and matched to the protein sequence, and each codon is aligned to its residue.
Superposition: Equivalent Cα atoms are paired through the SIFTS mapping or a sequence alignment, and the model is fitted by least-squares superposition.
Oligomer assembly: Foldseek searches the biological assemblies of the Protein Data Bank for templates, and a sequence search of the Protein Data Bank provides an alternative source. Each subunit is fitted onto the corresponding template chain, and steric clashes are relieved geometrically.
Contacts and packing: Residues are in contact when their heavy atoms lie within 4 Å. Packing is expressed as the number of Cα atoms within 10 Å of the residue.
Active-site distances: Distances are measured between the centres of the active-site residues annotated in UniProt, together with the closest pair of atoms. Interfaces and distances are measured in the structure as displayed; in a biological assembly, each symmetry copy of a chain is treated as a separate chain.
Substitution effects: DDMut estimates the change in folding stability and DDMut-PPI estimates the change in binding affinity, both in kcal/mol.

Interpretation of results

Predicted models are accompanied by confidence measures: pLDDT for each residue, predicted aligned error for residue pairs, and ipTM for interfaces. These measures guide the interpretation of each model and interface.

ΔΔG values are computational estimates with a reported error of 1.3 to 1.4 kcal/mol.

Citation

Please cite the data resources and prediction methods that contribute to your analysis.

Software and licences

The three-dimensional viewer is PDBe Mol*, distributed under the Apache License 2.0. The name is set in Cormorant Garamond and the headline in Cinzel, both distributed under the SIL Open Font License. This is Mohona version 0.15.4.

Help

User guide

Each section describes one task. Within the application, the information symbol beside a heading opens a description of that section.

Retrieving a structure

Type a PDB identifier, a UniProt accession, or a protein or gene name in the search field. Matching entries are listed beneath the field as you type.
Select an entry from the list, or press Enter to open the highlighted entry.
The Auto mode opens a PDB identifier as an experimental entry, and an accession or a name as an AlphaFold model.
The PDB and AlphaFold modes restrict the search to one resource.
The Ligand mode retrieves PDB entries that contain a given compound, identified by name, PDB ligand code or PubChem CID.
Select Open file in the navigation bar to load a structure in mmCIF or PDB format from your computer.

Sequence and structure

The Sequence tab displays the sequence of the selected chain, with secondary structure or pLDDT indicated beneath each residue.
Move the pointer over a residue to highlight it in the structure, and select a residue to centre the view on it.
Hold Shift and select a second residue to mark the range between the two.
The search field accepts a residue number or a sequence motif, for example N[^P][ST].
Select Protein + DNA to display each residue with its codon from the coding sequence.
Select FASTA or Gene FASTA to export the protein sequence or the coding sequence.

Molecular visualisation

The toolbar beneath the structure sets the representation and the colour scheme.
The toolbar also displays ligands and water molecules, rotates the model, restores the initial view and clears the selection.
For PDB entries, the toolbar switches between the biological assembly and the asymmetric unit.
Drag the divider between the two panels to adjust their widths.
The final button opens the control panels of the Mol* viewer.

Model confidence

The Confidence tab of an AlphaFold model reports the mean pLDDT and the proportion of residues in each confidence band.
A profile shows pLDDT along the sequence.
The predicted aligned error is displayed as a matrix. Drag across the matrix to colour the two corresponding residue ranges in the structure.
pLDDT values of 90 and above indicate very high confidence, and values from 70 to 90 indicate a confident backbone prediction.

Structural superposition

The Superpose tab lists the AlphaFold models of the proteins in a PDB entry, or the PDB entries of the protein in an AlphaFold model.
Select an entry from the list, or enter another identifier, to superpose the two structures.
Mohona reports the root-mean-square deviation over the paired Cα atoms and the deviation along the sequence.
The fit is recalculated for a selected residue range, and the superposed model is available for download.

Oligomer assembly

Open an AlphaFold model and select the Oligomer tab.
Select Find templates to search the biological assemblies of the Protein Data Bank with Foldseek. A sequence search of the Protein Data Bank supplies the templates when Foldseek is unavailable.
Use the search field above the list to filter templates by PDB identifier, title, organism or oligomeric state.
Alternatively, enter the PDB identifier of a template of your choice, with the assembly number after a hyphen, for example 5BS8-2.
Mohona identifies the corresponding protein from the same organism for each partner chain. Review the assembly plan and select Build oligomer.
The assembled model supports clash relief, comparison with the template and download of the coordinates.

Prediction of complexes

The Predict tab of an assembled model and the Metabolon tab prepare input files for AlphaFold Server, AlphaFold 3, Boltz-2 and OpenFold3.
Select a predictor and download the input file.
Run the prediction on AlphaFold Server, or on a computer on which the predictor is installed, with the command provided.
The installation scripts verify the operating system, graphics processor and available storage before each installation step.

Pathways and interaction partners

The Pathways tab reports the catalytic reaction, cofactors, pathway and pathway step of the selected protein.
Active-site and binding-site residues are listed. Select a residue to centre the view on it.
Proteins of the same organism that participate in the same pathway are listed with their pathway steps.
Interaction partners from STRING are listed with their scores and supporting evidence. Partners with evidence of membership in the same complex are marked as physical.
Select Open to load a partner, or Add to include it in the metabolon set.

Metabolon modelling

Assemble the set of enzymes with the Add buttons in the Pathways tab, or by UniProt accession in the Metabolon tab.
Add ligands and cofactors by PDB ligand code, and specify the number of copies of each component.
Review the evidence for each pair of enzymes: shared metabolites, consecutive pathway steps and STRING scores.
Download a single input file for the complete set and run the prediction.
Open the predicted model with Open file, and add the confidence file produced by the predictor.
The report lists the chains in contact, the ipTM and predicted aligned error of each interface, and the distances between active sites.
The same report is produced for the structure in the viewer, at the top of the Metabolon tab. For a PDB entry it covers the biological assembly or the asymmetric unit, as displayed.

Mutational analysis

In the Mutate tab, enter a substitution such as A91V, or select a residue and the replacement amino acid. In a complex, choose the chain from the list or name it first, as in B:A91V.
Each substitution is described by the underlying nucleotide change, the change in side-chain volume, hydropathy and charge, and the packing of the position.
Ligands, ions and neighbouring chains within 4 Å of the side chain are listed, together with the mutagenesis and variant records in UniProt.
For a PDB entry displayed as a biological assembly, the neighbours include the symmetry copies of the chains, ligands and ions, and the assembly is the structure sent for prediction.
Select Run DDMut to obtain the predicted change in folding stability from DDMut and the predicted change in binding affinity between chains from DDMut-PPI.
Negative ΔΔG values indicate destabilisation or reduced affinity. The reported error of these predictors is 1.3 to 1.4 kcal/mol.
The analysis sends the coordinates and the substitution codes to the DDMut servers at the University of Queensland.

Conventions

Residue numbers follow the numbering displayed in the Sequence tab.
Contacts in PDB entries are calculated for the structure as displayed: the biological assembly or the asymmetric unit.
Each tab presents its information in cards. The numbered cards of a workflow are completed in order.
Organism names are italicised throughout.
The information symbol beside a heading opens a description of that section, and the menu in the title band provides links and downloads.
The metabolon set, recent entries and display theme are retained in the browser between sessions.

Mohona: Where sequence, structure and pathway meet

Start with an example

Capabilities

Recently opened

Capabilities

Name and logo

Intended use

Data resources

Data handling

Structure prediction

Methods

Interpretation of results

Citation

Software and licences

Retrieving a structure

Sequence and structure

Molecular visualisation

Model confidence

Structural superposition

Oligomer assembly

Prediction of complexes

Pathways and interaction partners

Metabolon modelling

Mutational analysis

Conventions