KampoDB: Help

How to use KampoDB?

Part 1: Natural medicine list

In the "Search function" section, the user can input a natural medicine name (e.g., "kakkonto") as a query. Clicking on the search button, the user can obtain the corresponding information on Kampo medicines, crude drugs, constituent compounds, and target proteins. Note that compound IDs correspond to KNApSAcK IDs and protein IDs correspond to KEGG GENES IDs.

In the "Hierarchical relationship" section, the user can see a global classification of Kampo medicines, crude drugs, constituent compounds, and target proteins in a hierarchical manner. Note that each Kampo medicine consists of multiple crude drugs, each crude drug consists of multiple compounds, and each constituent compound is supposed to interact with its target proteins.

The hierarchy is organized in the following manner:

  • 1st layer: Kampo medicine (consisting of multiple crude drugs)
  • 2nd layer: crude drug (consisting of multiple constituent compounds)
  • 3rd layer: constituent compound (supposed to interact with its target proteins)
  • 4th layer: target protein
Please click on the "+" button for the lower layer (toward target protein), or click on the "-" button for the higher layer (toward Kampo medicine).

Part 2: Functional analysis

In the "Search function" section, the user can input a natural medicine name (e.g., "kakkonto") as a query. Clicking on the search button, the user can obtain the corresponding information on Kampo medicines, crude drugs, constituent compounds, and target proteins. Note that compound IDs correspond to KNApSAcK IDs and protein IDs correspond to KEGG GENES IDs.

In the "Enrichment analysis results" section, the user can see the summary of functional analysis for target proteins of the corresponding natural medicines, which provides pathway annotations and functional annotations of target proteins. Visualization of the results at different layer levels enables the user to see the mode-of-action information in a hierarchical manner within a natural medicine classification.

The hierarchy is organized in the following manner:

  • 1st layer: Kampo medicine (consisting of multiple crude drugs)
  • 2nd layer: crude drug (consisting of multiple constituent compounds)
  • 3rd layer: constituent compound (supposed to interact with its target proteins)
Please click on the "+" button for the lower layer (toward constituent compound), or click on the "-" button for the higher layer (toward Kampo medicine).

Please select one option from the following four categories and click on the corresponding button:

  1. Pathway: biological pathways in KEGG PATHWAY
  2. Brite: protein classifications in KEGG BRITE
  3. Process: biological process terms in GO
  4. Function: molecular function terms in GO

For example, in the case of "Pathway", the output is the list of pathway names with high enrichment ratio scores and low p-values. The enrichment ratio is defined as the ratio of the number of the associated target proteins to the number of all proteins in each pathway. The p-value is the false-discovery rate-corrected p-value for a hypergenometric test.

Part 3: Target prediction

Target proteins of constituent compounds are predicted by two computational methods: docking simulation and machine learning. In the docking simulation, docking was performed for consistutent compounds with each human protein 3D structure. In the machine learning, supervised regression for evaluating ligand structure similarity was performed.

Constituent compound

As a first step, please select a compound from the drop-down list showing compound names. They were used in the exploratory research project of Institute of Natural Medicine, University of Toyama, Japan. Then, you can see the 3D chemical strucutre of the selected compound below. Links to other molecular databases (PCIDB, ChEMBL, DrugBank, CTD, and PubChem) are also shown.

Docking simulation

To see results of docking simulations, please select a protein. You can choose it from a drop-down list, in which first 500 proteins in the database are shown. You can also type an Entrez Gene ID directly in the input field below the drop-down list. After selecting a protein or a corresponding gene in either way, click the botton labeled as "see results".

In a new window or tab, the results for the selected ligand and protein pair are shown. The ligand is shown in a ball-and-stick model, and a domain of the protein is shown in a space-filling model. If you cannot see the ligand in the view panel, please rotate the models in the view panel or change the view style to "Cartoon" in the outside panel titled as "View Style". You can also switch among top 5 binding modes and protein domains.

All the results have been pre-computed by AutoDock Vina. Protein domain structure models are obtained from SAHG [SAHG (c) Motono Chie (The Molecular Profiling Research Center for Drug Discovery (molprof), The National Institute of Advanced Industrial Science and Technology (AIST)) licensed under CC Attribution-Share Alike 4.0 International].

Target proteins predicted by machine learning

The bottom table shows a list of proteins targetting the selected ligand that are newly predicted by a machine learning method on the basis of compound chemical structures and chemical-protein interactome data. All the proteins targetting the selected ligand with prediction score above 0.4 are shown. The column labels have the following meanings:

Prediction score
confidence in the prediction ranging from 0 to 1 (higher is better)
Protein
protein name
Definition
KEGG definition of the protein
Docking simulation
link to the docking simulation result of the protein and the selected ligand
Pathway
KEGG pathways to which the protein belongs
Disease
diseases to which the protein is related

Data statistics in KampoDB on August 2, 2019

categorynumber
Kampo medicines42
crude drugs54
constituent compounds1230
target proteins460
3D models of constituent compounds95
3D models of protein domains42 330
docking simulation results3 186 784

How to cite KampoDB?

Please acknowledge KampoDB in your publications by citing the following paper:

  • Ryusuke Sawada, Michio Iwata, Masahito Umezaki, Yoshihiko Usui, Toshikazu Kobayashi, Takaki Kubono, Shusaku Hayashi, Makoto Kadowaki & Yoshihiro Yamanishi: KampoDB, database of predicted targets and functional annotations of natural medicines, Scientific Reports, volume 8, Article number: 11216 (2018). [link]


HOME