Skip to content

Home

Jump to bottom
AlexanderMon edited this page Feb 1, 2023 · 13 revisions

RING-PyMOL Wiki. Here you can find all the information about how to use the plugin and how the calculations are performed. Install information are provided in the README.

Configuration

RING

  • Run locally - You can execute RING from a local instance or on a remote server using the RING webserver API. Windows users are constrained to use the API as the RING software works only in Linux and MacOS operative systems.
  • Rerun RING every time - If you modify the underlying structure is better to rerun RING to avoid inconsistencies when the structure is modified
  • RING executable - Path to the RING executable. By default, RING-PyMOL test a list of default paths, if it fails the field is empty
  • Edges - The options for the edge calculation

    • Multi: it computes multiple edges for a pair of nodes, filtering out triangular connections like the one shown in the image below. It retains the connection with the lower distance computed between the interacting atoms

    • One:Return only the most valuable connection between two residues

    • All: Return all the connections found

  • Sequence separation - The minimum sequence distance between two residues to be considered for an interaction
  • Distance thresholds - The maximum euclidean distance between two atoms (or pseudo-atoms) to be considered for an interaction

Visualization

This tab let you customize the width, transparency and colors of the edges visualized.

Tutorial

RING execution

The RING-PyMOL plugin provides a fixed section on the top and a number of different tabs (NODES, EDGES, CLUSTERING, CONFIGURATION, ABOUT). The top section allows to execute the RING calculation and set some filtering parameters.

The plugin can be tested with the following example:

  1. Fetch the multi-state structure PDB code: 2H9R from the PDB with the command fetch 2h9r
  2. Open the RING-PyMOL plugin. Plugin $\rightarrow$ Ring plugin or type in the command line ring_plugin
  3. On the plugin top bar the object 2h9r should be already selected. Now you can execute RING clicking Execute RING. This runs the RING executable (if present), or call the web server depending on your configuration
  4. Once the results are ready you should see lines (edges) connecting atoms and two new Compiled Graphic Objects (CGO) representing interacting residues/groups of atoms obj_nodes and the interactions obj_edges. Edge objects contain sub-objects obj_interactionType that group interactions of the same type

Interactions are shown state-by-state. If you navigate different states the interactions may change, depending on the conformational diversity of the ensemble.

Edge filtering

After the first execution of RING on a PyMOL object the Execute RING button becomes Show, indicating the calculation has been performed correctly. The Show button serves to update the viewer when a different filter is applied. Filters include:

  • Interactions - All, Interchain, Intrachain, Sidechain
  • Frequency - If the analyzed object has multiple states then you can filter the interactions by their frequency, setting a minimum and maximum frequency. Interactions that have a frequency that is not in that range will be filtered out from the visualization.

In the example below are shown only interactions appearing in at least 10% of the states and less than 90%.

NODES tab

Pairwise interaction plot

It generates a plot showing an interaction across states. The $x$ axis is the number of states, the $y$ axis represents the interaction distance. Different series represent different types of interaction. Before clicking on the button, you need to create a PyMOL selection of exactly two residues, and selecting it in the RING-PyMOL top bar.

Nodes interaction table

It opens a new window with a table with the interaction frequency for each pair of nodes. The rows of the table can be selected in order to create a new PyMOL selection (sele_rows) with the selected residues. To reset the selection you can delete the corresponding PyMOL element.

The table is useful to easily see if a residue is involved in a certain type of interaction and its frequency when the structure is multi-state.

Color nodes by interaction frequency

This option lets you color the residues of the selected object based on the frequency of interaction of a certain type. The coloring is done using a heatmap, where the residues with the highest frequency are colored in blue, while the residues with the lowest frequency are colored in white. The frequency is computed on the number of states that a residue present the selected type of interaction over the total number of states.

Select the interaction of interest and then click on Color nodes to color the nodes in the structure.

EDGES tab

Interaction plots

The following two graphs give a quick overview of the interactions

  • Chain interactions - Generates a graph where node are the structure chains and the edges are the interactions between the chains
  • Secondary structure interactions - Generates a graph showing the interactions occurring between secondary structure elements. Nodes are the secondary structure elements calculated by RING (DSSP), $\alpha$ helices and $\beta$ strands.

Probabilistic interchain residue contact map

Generates a heatmap representing the probability of interaction between residues of different chains (default). The probability is computed as the number of states in which the two residues are interacting over the total number of states. The user can select different types of interactions using the drop-down menu. all shows all interactions types and also includes intrachain interactions. The heatmap can be zoomed and panned using the controls in the top bar, this can be useful to concentrate the plot in a specific region (e.g. chain A interactions with chain B).

In the figure below the probabilistic interchain residue contact map of $\pi-\pi$ stack interactions

Pairwise interaction correlation analysis

Generates a table that summarizes correlations between contacts found in a multi-state structure. Positive correlation means two contacts are observed in the same states, negative correlation (anti-correlation) means they alternate. If states represent MD snapshots contacts are correlated over the time variable and correlation can be a proxy to study allosteric paths.

Table columns are:

  • First contact
  • Frequency of the first contact in the multi-state structure, by interaction type
  • Interaction type of the first and second contact, when the interaction type is ALL this means that all types are considered
  • Second contact
  • Frequency of the second contact
  • Pearson' correlation between the two contacts
  • Correlation P-value

The Pearson' correlation coefficient and P-value are calculated with SciPy

Filtering

Spurious contacts and fixed contacts, that would have limited significance, can be removed before creating the table by setting the frequency filter. Once the correlation matrix is produced more filters can be applied on the columns headers. Multiple additive filters can be set using the space as separator. For example if you want to contacts of chain A with a correlation of 0.8 for hydrogen bonds, you can write: A/ 0.8 HBOND

Selection & visualization

Rows can be selected (with multi-selection active, with ↑Shift or ctrl ). By clicking Visualize selected, selected edges will be created in the PyMOL interface, and two selections will be created containing the residues in edge_1 and edge_2 columns. This way you can highlight the correlating interactions, and continue downstream RING-PyMOL analyses, such as the residue pair interaction plot on two edges to confirm that the correlation or anti-correlation is significant.

CLUSTERING tab

When dealing with large multi-state structures like those derived from molecular dynamics simulations, it can be helpful to reduce the number of states to highlight contact and conformational patterns emerging during the simulation.

Hierarchical clustering

RING-PyMOL provides a hierarchical clustering based on all-vs-all RMSD distance calculated across all states $(n\times n)$. By default, RMSD is calculated superimposing only the $C\alpha$ atoms for a faster computation. Superimposition is calculated using the BioPython superimposer module. The clustering algorithm can be selected from those provide by the SciPy Hierarchy module here. The clustering is computed on demand, when you click on plot or Create object buttons. The latter generates/updates an object with the representative states of the clusters. Once the hierarchical clustering is computed, the hierarchy tree can be cut by RMSD or by number of clusters.

Clustering visualization

Two different plots can be produced with the results of the clustering, giving the user an idea of the distribution of the clusters, their densities and separations.

Hierarchical clusters - It shows the hierarchy tree of the clusters. In the x axis the labels are cluster representative -(# states in cluster), i.e. the state with the lowest sum of distances between all other states in the same cluster.

State clusters - it shows each state as a cell, different colors represent different clusters. This provides an overview of the conformational dynamic across the simulation. Hovering a cell shows the state number and the corresponding cluster.

In the example figure we set a RMSD cutoff of 3.5 Å generating 5 different clusters.

Clone this wiki locally