Welcome to the KVFinder-web service!

A web service for cavity detection and characterization in any type of biomolecular structure

Step 1. Choose input

Type of input:

Fetch from PDB

Upload PDB file

Browse...

PDB ID:

The cavity detection can be performed by uploading a PDB file from your computer or by fetching the Protein Data Bank using a PDB ID.

In both cases the input is checked for non-standard protein or nucleic residues, for example water (HOH), ligands and ions molecules.

By default, KVFinder-web removes all non-standard protein residues from the input. For instance, if a target pocket is occupied by a ligand, that ligand is removed before computing the cavity.

In specific situations, users may be interested in evaluating a portion of a cavity that is occupied by a ligand, water or a ions. In this case, users can use Include option to include the non-standard residue in the analysis.

Step 2. Choose run mode

Detect cavities in:

Whole structure (default)

Whole structure (customized)

Around target molecule

Around target residues

KVFinder-web parameters

Probe in (Å):

Probe out (Å):

Removal distance (Å):

Volume cutoff (Å³):

KVFinder-web parameters

Probe in (Å):

Probe out (Å):

Removal distance (Å):

Volume cutoff (Å³):

Ligand or molecule name:

KVFinder-web parameters

Probe in (Å):

Probe out (Å):

Ligand cutoff (Å):

Removal distance (Å):

Volume cutoff (Å³):

Target residues:

KVFinder-web parameters

Probe in (Å):

Probe out (Å):

Padding (Å):

Removal distance (Å):

Volume cutoff (Å³):

In this box you need to set the mode of your KVFinder run.

In Whole Structure mode with default parameters, a preset of KVFinder-web parameters will be used to detect cavities in the whole biomolecular structure.

In Whole Structure mode with customized parameters, you can modify the KVFinder-web parameters to detect cavities in the whole structure. A help text is available when the you moves the mouse over the parameter.

The cavity detection can be focused on the region of the structure occupied by a target molecule. This mode limits the search space around a target molecule.

The cavity detection can also be focused on specific regions and residues of the target structure. This mode explores closed regions with a custom box, which is drawn by selecting residues of the target structure.

Get latest results

Insert the job ID to get results

Here, you can insert a job ID to get or check your latest results.

Tutorial

Welcome to the KVFinder-web service tutorial! This section provides an overview of the features available through our publicly available KVFinder-web service at ( kvfinder-web.cnpem.br ).

In this tutorial, we will guide you through the four execution modes of KVFinder-web, which enable you to identify and characterize cavities in biomolecular structures. We will also show you how to retrieve results using job IDs.

For this tutorial, we will be using two biomolecular structures accessible via their PDB IDs:

Human Peroxisome Proliferator Activated Receptor Gamma ( 1PRG );
HIV-1 Protease ( 1HVR )

By the end of this tutorial, you should have a good understanding of how to use KVFinder-web for cavity detection and characterization in your own biomolecular structures.

Cavity Analysis

To start your cavity analysis pipeline, open the "Run Cavity Analysis" tab.

Next, load your target biomolecular structure into "Step 1. Choose Input" by selecting "Fetch from PDB" as the input type and typing the PDB ID in the text box. For example, to analyze the human Peroxisome Proliferator Activated Receptor Gamma (1PRG), type "1PRG" and click "Load".

Once the structure is loaded, you will have the option to keep non-standard residues (such as water, ligands, ions, etc). For the 1PRG example, we do not want to keep water molecules for further analysis, so uncheck the "HOH" checkbox.

Now, we can explore the four different execution modes, which are:

Whole structure with default parameters

This mode is designed to make a simple and fast whole structure detection. To use it, select "Whole structure (default)" in the "Step 2. Choose run mode".

To submit a job for cavity analysis, click on the "Submit the job" button.

After successfully submitting the job to KVFinder-web service, a window will appear with the Job ID . For instance, the Job ID of our submission is 12727934810399286938 . To check your submission, click on "Check Results" button.

After completion, the "Status" field will change to "completed" and turn green, indicating that the job has finished processing. Then, the following three buttons will appear, along with a table showing cavity characterizations:

Download Structures: Download a ZIP file with PDB-formatted files of the input and the cavities;

Download Results: Download a TOML-formatted file with characterizations (volume, area, average depth, maximum depth, average hydropathy and surrounding residues) of the detected cavities and the parameters used in the cavity analysis;

View: Enable an interactive window to visualize the cavities with the target biomolecular structure.

Additionally, you can select a specific cavity to visualize in the interactive window by clicking on the corresponding name in the "Select cavity" drop menu. For instance, we select 'KAC' in the "Select cavity" drop menu.

You can also show the residues surrounding the selected cavities by clicking on the "Interface AA" check button.

Whole structure with customized parameters

To use customized parameters, select "Whole structure (customized)" in the 'Step 2. Choose run mode".

The KVFinder-web service allows users to customize detection parameters for the parKVFinder software. The parameters and their impact on cavity detection are explained in detail in the "Help" tab. To summarize, 'Probe In' defines the inner limits of cavities, 'Probe Out' defines the outer limits of cavities, 'Removal Distance' defines the length to be removed from the cavity-bulk boundary, and 'Volume Cutoff' excludes cavities with volumes smaller than the limit.

To demonstrate the effect of changing these parameters, let's vary "Volume Cutoff", "Probe Out", and "Removal Distance":

Adjusting Volume Cutoff: By setting "Volume Cutoff" to 500 Å³, smaller cavities are excluded from detection. After submitting the job and viewing the results using the "View" button, we can see that fewer cavities are identified compared to the detection with default parameters.

Adjusting Probe Out: By increasing the "Probe Out" size to 8.0 Å and keeping "Volume Cutoff" at 500 Å³, the detected cavities have a higher boundary than those detected with a 4.0 Å probe. Therefore, by increasing the size of "Probe Out", the cavity boundary is also raised.

Adjusting Removal Distance: Besides adjusting the "Probe Out" size, we can also adjust the "Removal Distance" to change the cavity boundary. By setting "Removal Distance" to 1.2 Å and keeping the size of "Probe Out" at 4.0 Å and "Volume Cutoff" at 500 Å³, the detected cavities have a higher boundary than those detected with a 2.4 Å "Removal Distance". Therefore, by decreasing the "Removal Distance", the cavity boundary is also raised.

Therefore, changing the cavity boundary by varying "Probe Out" and "Removal Distance" also affects cavity segregation.

Steered detection of cavities is another important feature of parKVFinder. To demonstrate this, we can analyze the HIV-1 protease structure (PDB ID: 1HVR) instead of the PPAR-gamma structure (PDB ID: 1PRG). To do this, we simply replace "1PRG" with "1HVR" in "Step 1" and click the "Load" button again.

Upon successfully loading the structure, we can also exclude specific molecules, such as the HIV-1 protease inhibitor XK2, by not checking the corresponding checkbox.

Then, we continue our tutorial by illustrating two distinct methods of cavity segmentation:

Around target molecules

In this section, we will explore a method of cavity segmentation that limits the search around a specific structure. We will use this method to detect the cavities of HIV-1 protease while focusing on its inhibitor.

To begin, select "Around target molecule" in "Step 2. Choose run mode". Next, select the "XK2" option in the Ligand or molecule name selection box. Then, submit the job to the KVFinder-web service.

Once the job is completed, click on the "View" button to display the job results.

Around target residues

This method allows us to limit the search around a closed region with a custom box, which is drawn based on a selection of residues and a padding. Let's detect the cavities of HIV-1 protease again, but this time limit the search space around residues "27_A; 27_B; 34_A; 34_B; 51_A; and 51_B;" and set the padding to 0.5 Å. Then, submit the job to KVFinder-web service.

Once the job is completed, view the job results by clicking on the "View" button.

Submitting an already sent job

When a job has already been submitted to the KVFinder-web service and is still available, the "Status" field will already be shown as "completed".

Retrieving a Job by ID

Users can easily share their job results with colleagues by providing them with the job ID. To retrieve a job, navigate to the "Retrieve results" page and enter the job ID in the "Insert the job ID to get results" text box. Then click on the "Get results" button. For instace, "12727934810399286938" (ID of our first job in this tutorial)

If the job is still available on the KVFinder-web service, a field will appear with the Job ID Status. After the job is completed, the "Status" field will change to "completed" and turn green. At this point, three buttons and a table displaying cavity characterizations will appear on the screen. To view the job results, simply click on the "View" button.

If the job does not exist or is no longer available on the KVFinder-web service, an error message will appear in the "Status" field.

KVFinder-web help page

Overview

KVFinder-web is an open-source web-based application of an updated version of parKVFinder software (v1.2.0) for cavity detection and characterization of any type of biomolecular structure, including but not limited to proteins and nucleic acids. KVFinder-web uses a geometrical grid-and-sphere based method with a dual-probe system to efficiently detect cavities. The web-based application also provides a comprehensive characterization that includes spatial, depth, constitutional, and hydropathy information. For a more detailed explanation, please refer Oliveira et al. 2014 , Guerra et al. 2020 and Guerra et al. 2021 .

Detection parameters

To detect cavities in biomolecular structures, KVFinder-web uses the following detection parameters:

Probe In (Å): A smaller probe that rolls around the target biomolecule and defines its surface. Typically, this is set to the size of a water molecule (1.4 Å).
Probe Out (Å): A larger probe that rolls around the target biomolecule. Users can adjust the size of the probe based on the characteristics of the target structure.
Removal Distance (Å): A length that is removed from the boundary between the cavity and bulk (solvent) region.
Volume Cutoff (Å³): A cavity volume filter to exclude cavities with smaller volumes than this limit. These smaller cavities are typically not relevant for function.

A schematic representation of cavity detection algorithm in KVFinder-web is presented below. In panel (A), a biomolecular structure X is inserted into a grid. In panel (B), Probe In (Pi) scans the surface of the structure, by translating over the grid points (orange). In panel (C), Probe Out (Po) scans the accessible points in blue. In panel (D), the cavity points (light grey) are defined as the difference between the probes' accessible points. The points not reached by Pi (dark grey) define the Solvent Excluded Surface (SES). In panel (E), a removal distance routine is applied to remove cavity points within a given distance from the cavity-bulk boundary (red line).

The impact of parameters on cavity detection

Two important parameters for cavity detection in KVFinder-web are Probe Out and Removal Distance, which directly impact execution time in KVFinder-web service.

Probe out: is a larger probe that systematically translates in a Cartesian grid around the target biomolecule, defining the boundary between the cavity and the bulk (solvent) due to the restricted access to the empty space within the biomolecule. Thus, greater 'Probe Out' sizes tend to reduce the degree of accessibility of the molecular surface created and ultimately, increase the elapsed time to perform calculations in KVFinder-web service.
Removal Distance: removes cavity points within a given length from the defined cavity-bulk boundary. Thus, reducing the 'Removal Distance' removes fewer points from the boundary, which helps to segregate sub-pockets and/or detect superficial cavities.

Apart from these parameters, the execution time of KVFinder-web increases linearly with the number of atoms of the target biomolecule inside the 3D grid. However, users can control the number of atoms by defining a custom search space around the ligand/molecule or residues, which can reduce the execution time.

Cavity naming convention

The naming convention for cavities in KVFinder suite (parKVFinder, pyKVFinder and KVFinder-web) is based on the integer label assigned to each cavity in the 3D grid. The first cavity identified in the grid is labeled as KAA, the second as KAB, the third as KAC, and so on.

The integer labels are assigned using a DFS clustering algorithm, which identifies cavities in the order they are found in the 3D grid. The integer labels used in the algorithm are:

-1: bulk (solvent);
0: biomolecule;
1: empty space (cavities that do not reach the volume cutoff);
>1: cavities.

Depth characterization

Depth characterization is a method used to identify the degree of burial of the binding site, with the largest and deepest cavity often corresponding to the active site in an enzymatic protein. This information can be helpful in identifying the active site throughout the molecular surface.

A schematic represention of the depth characterization is shown below. To perform depth characterization, a spatial filter is used to evaluate whether a cavity point (black) is a direct neighbor (red) of a bulk point, which assigns it as a boundary point (shown in panel A of the schematic). The distance between the cavity point (black) and the boundary points (red) is then calculated (shown in panel B of the schematic), and the shortest distance (blue line) is considered the depth of the cavity point. By calculating the depth of all cavity points, the maximum and average depths can be determined for each detected cavity.

Source: Guerra et al. (2021) . License: CC BY 4.0 .

Hydropathy characterization

Hydropathy characterization provides valuable insights into the water attractiveness and types of interactions at the interface of the binding site.

A schematic represention of hydropathy characterization is shown below. The Eisenberg & Weiss hydrophobicity scale ranges from -1.42 (highly hydrophobic) to 2.6 (highly hydrophilic), and projects the value of the amino acid nearest to a target surface point. Panel A displays the hydrophobicity of each amino acid. The distance between a surface point (black) and the atoms (carbon: green; oxygen: red; nitrogen: blue; sulfur: yellow) of the amino acids (methionine: left; valine: right) are calculated, and the residue of the atom with shortest distance (red line) maps its hydrophobicity value onto it (shown in panel B).

Modified from: Guerra et al. (2021) . License: CC BY 4.0 .

KVFinder-web service requirements and limitations

The KVFinder-web service has a Job timeout of 30 minutes (the maximum time that an accepted job could run on the server), completed jobs will be available on the web service up to 1 day after completion, and the maximum payload (maximum size of the JSON) of the data sent to the KVFinder-web service is 1 MB.

Further, the KVFinder-web service has some limitations, compared to a local installation of parKVFinder, that are:

The Probe In and Probe Out sizes must be smaller than 5 and 50 A, respectively, to avoid unnecessary time-consuming jobs.
The Removal distance is limited and must be smaller than 10 A to avoid unnecessary time-consuming jobs.
The grid spacing is preset to 0.6 A to avoid unnecessary time-consuming jobs.
Cavity representation will be always filtered (cavity files will consume less space on the web service).

This website is free and open to all users and there is no login requirement.

About KVFinder-web

The KVFinder suite

KVFinder suite comprises a toolkit of cavity detection and characterization software with different scopes, that are:

KVFinder-web

KVFinder-web is an open-source web-based application of an updated version of parKVFinder software (v1.2.0) for cavity detection and characterization of any type of biomolecular structure, including but not limited to proteins and nucleic acids. KVFinder-web implements a geometrical grid-and-sphere based method to detect cavities using a dual-probe system. KVFinder-web provides a comprehensive characterization of biomolecular cavities that includes spatial, depth, constitutional, and hydropathy information.

The KVFinder-web has two independent components:

a RESTful web service: KVFinder-web service
a graphical web portal: KVFinder-web portal

Citing

If you use KVFinder-web , please cite:

João V.S. Guerra, Helder V. Ribeiro-Filho, José G.C. Pereira, and Paulo S. Lopes-de-oliveira. KVFinder-web: a web-based application for detecting and characterizing biomolecular cavities. Nucleic Acids Research, May 2023. URL: https://doi.org/10.1093/nar/gkad324 , doi:10.1016/10.1093/nar/gkad324.
João Victor Da Silva Guerra, Helder Veras Ribeiro Filho, Leandro Oliveira Bortot, Rodrigo Vargas Honorato, José Geraldo De Carvalho Pereira, and Paulo Sérgio Lopes-de-oliveira. ParKVFinder: a thread-level parallel approach in biomolecular cavity detection. SoftwareX, 12:100606, July 2020. URL: https://doi.org/10.1016/j.softx.2020.100606 , doi:10.1016/j.softx.2020.100606.

If you use depth and hydropathy characterization, please also cite:

João Victor da Silva Guerra, Helder Veras Ribeiro-Filho, Gabriel Ernesto Jara, Leandro Oliveira Bortot, José Geraldo de Carvalho Pereira, and Paulo Sérgio Lopes-de-Oliveira. pyKVFinder: an efficient and integrable python package for biomolecular cavity detection and characterization in data science. BMC Bioinformatics, December 2021. URL: https://doi.org/10.1186/s12859-021-04519-4 , doi:10.1186/s12859-021-04519-4.

parKVFinder

Parallel KVFinder (parKVFinder) is a powerful open-source software designed to detect and comprehensively characterize biomolecular cavities of any type. It provides detailed information on the spatial, depth, constitutional, and hydropathy characteristics of each cavity. The cavity detection process relies on a customizable set of intuitive parameters that users can interact with via a graphical user interface (GUI) or a command-line interface.

The spatial characterization of each cavity includes information on its shape, volume, and surface area. The depth characterization determines the depth of each cavity point, shown in the B-factors, and calculates the average and maximum depth per cavity. The constitutional characterization identifies the amino acids that form the identified cavities. Lastly, the hydropathy characterization maps the Eisenberg & Weiss hydrophobicity scale onto the surface points of each cavity, shown in the Q-factor, and estimates the average hydropathy for each cavity.

For more information on parKVFinder , please refer to: [2]

pyKVFinder

Cavity analysis can be a data-intensive process that requires efficient scripting routines built on easily manipulated data structures. To address this need, we developed Python-C Parallel KVFinder (pyKVFinder), an open-source (GPL v3.0) Python package to detect and characterize cavities in biomolecular structures for data science and automated pipelines.

pyKVFinder uses a dual-probe algorithm to scan the biomolecular structure inserted into a regular 3D grid stored as an N-dimensional array (ndarray). The software detects cavities and provides comprehensive information on their spatial, depth, constitutional, and hydropathy characteristics.

pyKVFinder is specifically designed to facilitate biostructural data analysis in the Python ecosystem and is an essential building block for data science and drug design applications. It is an excellent choice for automated pipelines that require efficient and fast processing of large datasets. With pyKVFinder, researchers and developers can easily manipulate and process cavity data structures to extract valuable insights and develop new drugs or biomolecular applications.

For more information on pyKVFinder , please refer to: [3]

Troubleshooting

If you encounter any issues or have requests related to the KVFinder-web service, please don't hesitate to inform us so that we can make improvements. We manage our project through GitHub and you can use our Issues page to either:

report a bug ;
request a new feature .

Before submitting your report or request, please make sure it has not already been reported by using the search function to look for keywords.

Development Team

The following team develops and maintains KVFinder-web:

João Victor da Silva Guerra
Helder Veras Ribeiro-Filho
José Geraldo de Carvalho Pereira
Paulo Sergio Lopes-de-Oliveira (Principal Investigator)

The team belongs to the Computational Biology Laboratory ( LBC ) at the Brazilian Biosciences National Laboratory ( LNBio ) in the Brazilian Center for Research in Energy and Materials ( CNPEM )

LBC Staff

Gabriel Ernesto Jara
Helder Veras Ribeiro Filho
João Victor da Silva Guerra
José Geraldo de Carvalho Pereira
Leandro Oliveira Bortot
Paulo Sergio Lopes de Oliveira (Principal Investigator)

Contact

If you have any questions, inquiries, or wish to contribute to the KVFinder suite, please feel free to contact us. You can reach us via email at joao.guerra@lnbio.cnpem.br or paulo.oliveira@lnbio.cnpem.br .

Funding

The KVFinder-web software was supported by Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP) [Grant Number 2018/00629-0], Brazilian Biosciences National Laboratory (LNBio) and Brazilian Center for Research in Energy and Materials (CNPEM).

License

The KVFinder-web software is licensed under the terms of the Apache-2.0 License and is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the Apache-2.0 License for more details.