Mapping of the Available Chemical Space versus the Chemical Universe of Lead‐Like Compounds

Abstract

This is, to our knowledge, the most comprehensive analysis to date based on generative topographic mapping (GTM) of fragment-like chemical space (40 million molecules with no more than 17 heavy atoms, both from the theoretically enumerated GDB-17 and real-world PubChem/ChEMBL databases). The challenge was to prove that a robust map of fragment-like chemical space can actually be built, in spite of a limited (≪10 ) maximal number of compounds ("frame set") usable for fitting the GTM manifold. An evolutionary map building strategy has been updated with a "coverage check" step, which discards manifolds failing to accommodate compounds outside the frame set. The evolved map has a good propensity to separate actives from inactives for more than 20 external structure-activity sets. It was proven to properly accommodate the entire collection of 40 m compounds. Next, it served as a library comparison tool to highlight biases of real-world molecules (PubChem and ChEMBL) versus the universe of all possible species represented by FDB-17, a fragment-like subset of GDB-17 containing 10 million molecules. Specific patterns, proper to some libraries and absent from others (diversity holes), were highlighted.

The Symbiotic Relationship Between Drug Discovery and Organic Chemistry

“…10 63 molecules with 30 atoms or less);e ven if only the subsection of the lead-like part is considered, it is still impossible to comprehend by any synthetic and even most virtuals creening methods. [33] To sample the chemical space in an efficient manner, one might select molecules that are as diversea spossible; indeed, diversity is one of the key features considered in compoundl ibrary design.A lthough understandablea tt he intuitive level, molecular diversity,a sw ella st he related concept of moleculars imilarity,i sd ifficult to define in quantitative terms. Primarily,t his is relatedt ot he fact that these concepts are not uniquelyd efined and cannot be inherently "objective."…”

Diversifying chemical libraries with generative topographic mapping

“…The GTM method relates the data points positions in the initial N-dimensional space and in the latent 2D space. The GTM algorithm is described in a range of publications [10][11][12]18]. Briefly speaking, GTM injects a 2D hypersurface (manifold) into a multidimensional data space populated by a set of representative items (the Frame Set, FS).…”

“…In our early study [12], frame set compounds were randomly selected from large chemical libraries. Here, a FS containing 25K AMS compounds of controlled diversity (featuring no two compounds more similar than a given threshold) was prepared.…”

“…Recently we demonstrated that GTM represent an efficient tool for comparison of large chemical libraries FDB-17 and PubChem-17 [12]. Hierarchical zooming technique [13] was successfully applied in [12] in order to analyze the chemotypes of molecules populated selected zones and to highlight the scaffolds present exclusively in FDB-17.…”

Cheminformatics in Drug Discovery, an Industrial Perspective

“…A general lecture in cheminformatics must by necessity be selective and will not be able to cover all aspects of the field. An important are that won't be discussed is visualization of the chemical space, which is an active research area ,. Besides traditional cheminformatics tasks as outlined above, the use of machine learning in cheminformatics has become a very hot topic and will be extensively discussed in the lecture…”