SwissCellPrediction

How to use SwissCellPrediction



Welcome to SwissCellPrediction documentation providing technical explanations and annotated screen captures to properly use the Web interface.
To perform a reverse virtual screening with SwissCellPrediction, you first need to input a query molecule.

Input query molecule

The query molecule can be inputted either by (1) writing its SMILES directly in the left-hand text box, or (2) using the molecular editor on the right, which enables user-friendly drawing of chemical structures. The molecular editor also allows to import chemical structures (a) from a local file or a remote file through the network, or (b) by its chemical or common name. The sketcher window is synchronized with the SMILES text box and has copy/paste features, providing an overall easy way to modify chemical structures. The user can select some example molecules from the drop-down menu. A “clear” button is available to delete any structure already entered. Once the query molecule set, the user can launch the computation by clicking on the red button: “Predict cell lines”.

Result of ligand-based reverse screening

Cell lines are ranked according to their probability of sensitivity (green bars in the main result table) for the query molecule. This probabilitiy value is a logistic regression score from the 2D/3D similarity with all known actives on the cell line. The main table also gives information about the predicted cell lines as species, tissues and associated disease. Links to more detailed descriptions are provided by clicking on the respective ChEMBL and Cellosaurus IDs (directing to the webportal of the database of origin).The last column shows the numbers of known actives that reached a similarity score of at least 0.4 in 2D (Tanimoto coefficient on FP2 fingerprints) and 0.8 in 3D (Manhattan-based similarity on ES5D vectors). The main table can be sorted by clicking on any column header. A text search box allows to filter the results.

The pie chart displayed in the top-right box shows the repartition of the primary diseases (Cancer Cell Line Encyclopedia). The top-left box contains the structure of the query molecule and interoperability icons to send it to other SwissDrugDesign webtools developed by the Molecular Modeling Group of the SIB Swiss Institute of Bioinformatics:
  • SwissTargetPrediction (target icon for prediction of protein targets)
  • SwissSimilarity (twins icon for ligand-based virtual screening)
  • SwissADME (pill icon for calculation of parameters relevant for physchem, ADME and medicinal chemistry)
  • SwissBioisostere (hexagon icon for replacement of molecular fragment)
  • SwissParam (sigma icon for parametrizations of molecular docking) are reachable this way
  • SwissCellPrediction (cell icon for prediction of cell line targets)
The interoperability between SwissCellPrediction and SwissTargetPrediction is particularly interesting, as the complementarity to predict both the cell lines and the proteins targets for the same query molecule can help deciphering mechanisms of cytotoxicity.

Export options are accessible through dedicated red icons. The user can download the main result table in different formats (PDF, CSV or Excel), copy it to the clipboard or print it directly . The the number of predicted cell lines to display can be adjusted to 15 (default), 25, 50 or all (up to 100).

Numbers in the last column of the main result table are link to other result pages that display the known actives for the corresponding predicted cell line.

Known actives on a predicted cancer cell line

The list of known actives on the corresponding predicted cell line, is ranked by decreasing (3D or 2D) similarity to the query molecule with access to the ChEMBL Compound Report Card by clicking any ChEMBL ID. An estimation of membrane permeation capacity is indicated. If the active was flagged as ubiquitous the responsible chemical pattern is given.
Another red icon just above the table allows the user to export the list of known actives in CSV format.

Case of ubiquitous query molecule

A query molecule is flagged as ubiquitous if it exhibits a chemical pattern associated to a cytotoxic mode of action leading to cell line non-specificity. In this case, SwissCellPrediction does not perform reverse virtual screening, because the anticancer activity is then assumed to be due to this substructure, rather than to molecular similarity with a known active. Therefore, the user is redirected to a different first output page, displaying a table with a list of cell lines experimentally validated as sensitive against some other compounds harboring the same ubiquitous flags as the query molecule. No similarity nor probability is calculated.

Other information

If you need help with the command-line, you can find documentation here.

If you need more help on the web interface, please have a look at the video tutorials.

Answers to frequently asked questions can be found in the FAQ.