Improving Drug-Likeness With Minimal Impact on Chemical Diversity

    Background

    There are two major philosophies in Al-driven structure generation in drug discovery.

    One common approach is to constrain models to “synthesizable” chemistry upfront. However, this narrows the chemical space to be explored, leading Al to propose non-novel structures that medicinal chemists could readily design without utilizing Al.

    At Elix, our motto is to expand the imagination of medicinal chemists. Thus, we keep the explored chemical space broad and use medicinal chemistry knowledge to guide generation away from clearly undesirable chemistry via reusable filter sets manually curated by our internal medicinal chemists.

    Elix Filters in Elix Discovery™

    Elix Filters are Elix’s proprietary substructure filters designed to exclude molecules with undesirable substructures while preserving broad exploration. These filters are manually curated and continuously updated by multiple in-house medicinal chemists, drawing on accumulated wet-lab results and Al-enabled drug discovery experience. As a result, they are challenging to reproduce outside Elix without long-term, medicinal chemist-led curation.

    They can be incorporated directly into the objective used during structure generation rather than only screening final outputs.

    By applying the Elix Filters, users can uplift the quality of every generation loop by avoiding known “non-starters” and well-known liabilities.


    アプリケーションノート全文を入手するには
    以下フォームにご入力ください:

    所属組織名
    氏名
    メールアドレス

    プライバシーポリシーに同意して上記内容を送信します。

    このサイトはreCAPTCHAによって保護されており、Googleのプライバシーポリシー利用規約が適用されます。