CoChat

Chemically Advanced Template Search (CATS) for Scaffold‐Hopping and Prospective Target Prediction for ‘Orphan’ Molecules

Open access
Reutlinger, Michael, Koch, Christian P., Reker, Daniel, Todoroff, Nickolay, Schneider, Petra, Rodrigues, Tiago, Schneider, Gisbert · 2013 · Molecular Informatics
Cited by 166

The DOI matches its Crossref record; 53 of the 53 references it cites were found in a registry or an index.

Abstract

Drug discovery is driven by the identification of new chemical entities (NCEs).1,2 Virtual screening and de novo design techniques have been proven to serve this purpose, thereby complementing experimental biochemical and biological approaches.3 Still, it remains a matter of debate, which particular molecular representation and similarity index are preferable for a given drug target in order to identify appropriate NCEs with minimal synthetic and testing effort involved.4 Ligand-based chemical similarity approaches have been effectively applied to large-scale activity and target prediction for known drugs, some of the prominent methods being PASS developed by Poroikov et al.,5 the techniques conceived by Mestres and co-workers,6 and the similarity ensemble approach (SEA) implemented by the Shoichet group.7 Here, we compared several popular two-dimensional molecular representations for their ability to retrieve actives (enrichment potential) and chemotypes (scaffold-hopping potential) from a collection of druglike bioactive compounds. Subsequently the applied chemical advanced template search (CATS)8 was applied to predicting potential drug targets for a virtually assembled combinatorial compound library, from which we synthesized and successfully tested candidate compounds. The results demonstrate that CATS is not only suited for its intended purpose of NCE retrieval by scaffold-hopping,9 but also for reliable target profiling of 'orphan' virtual molecules.10 It thereby complements the suite of available validated tools for target prediction. A framework for retrospective evaluation of similarity searching runs with different molecular representations ('descriptors') was established on basis of the COBRA collection of druglike bioactive compounds,11 employing Euclidean distances for metric descriptors and the Tanimoto coefficient for fingerprint descriptors.12 COBRA contains 12 642 manually curated entries with 980 target protein subtype annotations. For 170 macromolecular drug targets with a minimum of 20 annotated active ligands per target, each compound annotated as 'active' was selected as a query in turn, and compared to all remaining compounds in the screening pool in terms of molecular descriptor similarity, finally yielding sorted results lists with the most similar or least distant pool compounds sorted to the top. Although there are large collections of bioactive compounds available in the public domain,13 we used the carefully compiled COBRA collection to i) reduce the risk of erroneous activity data and faulty compound structures,14 and ii) avoid redundancy with existing tools that are based on such public structure-activity data. In addition, we intend to probe the value of a comparably small but well curated reference compound pool for target prediction. We used a representative set of descriptors and fingerprints for benchmarking. 'Morgan' fingerprints, closely related to extended-connectivity fingerprints (ECFP), are based on radial assessment of non-predefined potentially infinite molecular fragments.17 The 'AtomPair' descriptor can be seen as a CATS predecessor merely denoting the occurrence of all pairs of atoms at a given topological distance.18 The 'MACCS' keys represent substructure-based fingerprints,19 and the 'RDkit' fingerprint implements a Daylight-like fingerprint based on hashed molecular subgraphs.20 Latter fingerprints and descriptors were calculated using the open-source software package RDkit.21 Finally, the 'MOE2D' descriptor consists of a standardized vector of physicochemical properties provided by the Molecular Operating Environment (v2011, Chemical Computing Group, Montreal). At this point, we analyzed two versions of CATS vectors, namely the originally described CATS18 and CATS2, which distinguishes lipophilic from aromatic atoms during typing, thereby resulting in more pharmacophore type pairs and consequently a higher dimensionality of the descriptor than CATS1, which lacks the aromatic atom type. For both descriptors we employed 'types scaling', which mitigates the potential dominance of prevalent pharmacophore feature types, and a maximal correlation distance of 10 bonds.22 An example of CATS descriptor calculation is presented in Figure 1. Principle of CATS descriptor calculation. The molecular structure (Step 1) is reduced to the molecular graph, and feature types are assigned (Step 2; L, lipophilic; R, aromatic; A, hydrogen-bond acceptor; D, hydrogen-bond donor). Then, atom pairs for all feature pairs are counted (Step 3), and the final descriptor values are scaled (Step 4). Here, the raw values were divided by the respective λ value (sum of atom type pair occurrences). Note that not all vertices in the molecular graph are considered 'pharmacophoric'. These possess no feature types. We employed the Receiver Operating Characteristic (ROC) related BEDROC score for actives-retrieval benchmarking.15 For our study, the alpha level of the BEDROC method was set to 160.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1002/minf.201200141

Cites

53 of 53 references verified0 citation errors · 0 not checkable

Statistics

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Cited by

See all 166

Metadata

Authors
Reutlinger, M., Koch, C. P., Reker, D., Todoroff, N., Schneider, P., Rodrigues, T., & Schneider, G.
Venue
Molecular Informatics
Publisher
Wiley
Published
2013-02
Volume
32
Issue
2
Pages
133-138
ISSN
  • 1868-1743 (ISSN-L)
  • 1868-1751
OpenAlex
W2061937956
Cited by
166
References
53
FWCI
16.29
Type
journal article
Language
en
Open access
hybrid

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.