Virtual screening


    Click here to access the Virtual Screening Web Server

    Here you may submit long term background jobs to :

    (A) search our collection of > 3 million registered commercially available and reference compounds (Maybridge, Zinc, QSAR reference sets from the literature) by substructure or virtually screen by similarity - based on various original descriptors - starting from one or several queries (consensus similarity screening is supported).
    Searching can be performed either with ChemAxon standard tools, or using our original and complementary fragment and pharmacophore ISIDA descriptors.
    While ISIDA fragments capture topological similarity, fuzzy pharmacophore triplets support "scaffold hopping" - retrieval of original skeletons carrying a same pharmacophore pattern as in the query compounds.
    Several similarity/substructure-based queries can be processed independently and resulting lists combined (Boolean operators).
    Please note that fuzzy pharmacophore fingerprints (FPT) are information-rich high-dimensional integer value vectors : their comparison is nowhere close in speed to binary fingerprint matching (count several hours for a complete FPT-based similarity screening of the data base - albeit access to partial results is quick, and you may disconnect any time and revisit the page later).

    (B) predict compound properties based on large-scale consensus models obtained by the Stochastic QSAR Sampler.
    In complement to the quick, in-line Predictor tool, these approaches rely on thousands of linear and non-linear equations for each prediction, and must be run as a background task.
    The methodology returns .csv files and/or .html result pages, regrouping the predicted property and the confidence level associated to each prediction.
    The latter is based on an analysis of the situation of the compound to predict with respect to the Applicability Domain of each individual equation, and is associated to explicative remarks.