Pith. sign in

REVIEW 3 major objections 6 minor 58 references

WelQrate: Defining the Gold Standard in Small Molecule Drug Discovery Benchmarking

T0 review · 3 major / 6 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read WelQrate makes the case that curated screening data, not just model architecture, determines whether drug-discovery benchmarks are trustworthy.

desk verdict Useful, carefully documented benchmark with an overclaimed 'gold standard' label and an unquantified inactive-label noise problem. read the letter →

arxiv 2411.09820 v1 pith:WWW42MF5 submitted 2024-11-14 cs.LG cs.AIq-bio.BM

Yunchao Liu , Ha Dong , Xin Wang , Rocco Moretti , Yu Wang , Zhaoqian Su , Jiawei Gu , Bobby Bodenheimer
show 3 more authors
Charles David Weaver Jens Meiler Tyler Derr
This is my paper · ORCID
classification cs.LGcs.AIq-bio.BM
keywords drugdiscoverybenchmarkingvirtualscreeninghigh-throughputdatasetcurationhierarchicalbioassaymolecularrepresentationlearningscaffoldsplitearlyenrichmentmetrics
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper argues that machine-learning models for small-molecule drug discovery are being compared on benchmarks with noisy labels, inconsistent chemical representations, and ad hoc splits, so rankings may not reflect real ability. To fix this, it proposes WelQrate: nine datasets across five therapeutic target classes, curated by hand-inspected hierarchies of primary, confirmatory, and counter screens from the public bioassay database, followed by filters for promiscuous, interfering, and non-druglike compounds. It wraps these datasets in a standardized evaluation protocol covering molecular formats, 3D conformations, early-enrichment metrics, and random versus scaffold splits. The benchmarking results show that dataset quality, featurization, and split type all shift model rankings, with a domain-expert descriptor baseline often beating deep learning models. If the curation is sound, the community gains a common testbed on which virtual-screening claims can be compared fairly.

What carries the argument

The load-bearing mechanism is the hierarchical curation pipeline. It organizes bioassays by level: a primary screen with a deliberately loose threshold, confirmatory screens that re-test putative actives, and counter screens that reject compounds with off-target or nonspecific activity; final actives are validated hits and final inactives come from primary-screen inactivity, with noted exceptions kept when follow-up readouts contradict. This pipeline is what converts raw screening data into labels the paper claims are clean and realistic. Around it, the framework adds standardized formats (isomeric SMILES and InChI, plus precomputed 2D and 3D graphs), early-enrichment metrics such as logAUC, BEDROC, EF100, and DCG100, and split protocols including nested-style random cross-validation and Bemis-Murcko scaffold splits. The central working assumption is that label quality, not model architecture alone, determines whether a benchmark ranking transfers to real screening.

What would settle it

Re-test a random sample of the final inactive compounds from one WelQrate dataset, say AID1798, in the confirmatory assay used there (AID1488) under dose-response conditions; if a substantial fraction, for example several percent, reproducibly show activity, the inactive labels are too optimistic and the clean-label premise fails. A cheaper version is to search the public bioassay records for compounds labeled inactive in WelQrate that later returned active readouts in related follow-up assays and count how often that happens.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central claim is that a benchmark built from hierarchically curated high-throughput screening data, rather than raw primary-screen readouts, can serve as a gold standard for small-molecule virtual screening. The curation works by tracing each compound through primary, confirmatory, and counter assays, keeping only actives that survive validation and only inactives from primary screens that were not contradicted by later readouts, then applying promiscuity, PAINS, druglikeness, and representation filters. Around this collection, the paper standardizes featurization, 3D conformation generation, evaluation metrics that reward early enrichment, and split schemes including scaffold splits. Its benchmarking experiments find that models improve with complexity under random splits, that a domain-expert descriptor with a simple classifier outperforms the neural models, that training on uncurated primary-screen data changes or reverses some comparisons, that predefined features beat one-hot features, and that all models struggle under scaffold splits. The paper takes these results as evidence that both data quality and evaluation design must be reported and standardized for meaningful model comparison.

Load-bearing premise

The benchmark's clean-label claim rests on the assumption that compounds labeled inactive, mostly from primary screens without confirmatory or counter-screen validation, really are inactive; if primary-screen misses are common, the inactive half of every dataset is noisier than the curation suggests.

Editorial extensions

If this is right

  • Model rankings on WelQrate become a more trustworthy basis for choosing virtual-screening methods, because labels have passed confirmation and counter-screening.
  • The strong performance of a simple model on domain-expert descriptors implies that architecture comparisons should include such a baseline and that featurization matters as much as model design.
  • Training on uncurated primary-screen data can invert or erase performance differences, so benchmark results that ignore curation are hard to interpret.
  • Scaffold splits expose distribution shift: all tested models lose accuracy, meaning claims of generalization need to be evaluated under scaffold rather than random splits.
  • Adopting standardized metrics that reward early enrichment aligns benchmark scores with the real workflow of buying or synthesizing only the top-ranked compounds.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Beyond the paper, the curation's weak point is the inactive set: most inactives are primary-screen misses never validated in confirmatory screens, so if primary-screen false negatives are frequent, the clean-label premise is weakened even though actives are well validated.
  • A testable extension is to re-run the benchmark after applying a stricter inactive definition, for example requiring inactivity in a confirmatory or counter screen, and measure how rankings shift.
  • The additional dose-response measurements available for three datasets invite a regression benchmark for potency prediction, which the paper mentions but does not develop.
  • If the gold-standard framing is adopted widely, the field's next problem becomes split and feature variation across benchmarks; WelQrate's fixed protocols make cross-paper comparisons possible only if researchers report the version and split used.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 6 minor

Summary. The paper introduces WelQrate, a benchmark suite for small-molecule drug discovery consisting of nine PubChem-derived datasets with a hierarchical curation pipeline, standardized data formats (SMILES, InChI, SDF, 2D/3D graphs), a proposed evaluation protocol (metrics, random and scaffold splits, adapted cross-validation), and extensive benchmarking of ten models across four research questions. The central claims are that the datasets have clean and reliable labels due to expert-designed curation, and that the framework provides a standardized, realistic basis for virtual-screening model comparison. The paper also argues that dataset quality, featurization, and split strategy materially affect model rankings, and recommends WelQrate as a new gold standard.

Significance. If the curation and evaluation claims hold, WelQrate would be a valuable community resource: the authors provide public data, curation code, and experimental scripts; the datasets are large and realistically imbalanced; the supplementary material documents the full curation hierarchy for each AID; and the benchmarking includes standard and domain-baseline models with error bars and hyperparameter tables. The inclusion of multiple realistic ranking metrics (logAUC, BEDROC, EF100, DCG100) and scaffold-split evaluation is a genuine strength. However, the significance is contingent on two load-bearing points: the reliability of the inactive labels, which are mostly unconfirmed primary-screen negatives, and the validity of the adapted cross-validation protocol. Because the paper's headline contribution is 'clean and reliable data labels,' the inactive-label issue directly affects the benchmark's core value proposition and must be addressed before the gold-standard claim is supportable.

major comments (3)
  1. [Sec. 3.2; Supp. A.2] The central claim of 'clean and reliable data labels' (Sec. 3) is not established for the majority class. For most datasets, the final inactive set is taken directly from primary-screen inactive readouts (e.g., AID1798, AID435034, AID1843, AID2258, AID2689, AID485290; see Supp. A.2), and the paper itself notes that primary HTS thresholds are deliberately loose to reduce false negatives (Sec. 3.2). The curation notes also document primary-inactive compounds later found active in confirmatory screens: AID435008 states that 'we found some inactive compounds in one screen that were active in the other'; AID2258 describes six primary-inactive compounds that were retested, with final active readouts causing their exclusion; AID488997 reports 17, 38, and 2 primary-inactive compounds tested in confirmatory screens, some of which were active. No dataset-level false-negative rate is reported anywhere, and Section 6 (Limitations) does not mention inactive-label noise. Since inactives constitute the vast majority of each dataset, unquantified false negatives can distort enrichment and ranking metrics, so the clean-label premise needs either explicit quantification using the available retest data or a substantially softened claim and a corresponding limitation statement.
  2. [Sec. 5.2; Fig. 4] The RQ2 conclusion that the results 'align with data-centric AI, highlighting the importance of dataset quality' is not supported for all model families: the 2D and 3D graph-based models trained on the less clean control data outperform models trained on WelQrate in logAUC[0.001,0.1] and BEDROC, while performing worse on EF100 and DCG100. The authors offer only an untested hypothesis about the 'range of top selected candidates.' As Fig. 4 averages across datasets and the effect is metric-dependent, the conclusion should either be restricted to the architectures and metrics where curation consistently helps (Naive, sequence-based, and Domain baselines) or be backed by a per-dataset, per-metric analysis that explains the reversal. As written, RQ2 does not provide a coherent demonstration that dataset quality improves model evaluation across the board.
  3. [Sec. 4.3; Supp. B.2] The adapted cross-validation protocol, in which the validation fold is fixed as the fold immediately preceding the test fold, is asserted to 'enhance computational efficiency without compromising robustness,' but no evidence is provided that this protocol yields estimates comparable to nested cross-validation or to standard k-fold cross-validation with proper hyperparameter tuning. Since all random-split results (RQ1-RQ3) rely on this protocol, the evaluation framework's reliability claim depends on this untested assumption. A small-scale comparison of the adapted protocol against nested cross-validation on one or two datasets would be sufficient to support the claim, or the assertion should be replaced with a more cautious statement.
minor comments (6)
  1. [Sec. 1] There is a typo in the bullet list: 'theurapeutic' should be 'therapeutic'; the author affiliation line also contains a stray 'Electical and Computer Engineering Dept„' with a formatting artifact.
  2. [Sec. 4.3] The text says 'a 3:1:1 training:validation ratio' but appears to mean a 3:1:1 train:validation:test ratio; please clarify the wording to avoid ambiguity about whether the test set is included.
  3. [Table 1] In the AID1843 row, the number of unique BM scaffolds is listed as '82,140C' with a stray 'C' suffix; this appears to be a typo.
  4. [Supp. B.1] The displayed BEDROC formula after 'calculated as:' appears garbled: the summation term is written as 'P n i=1 -eri/N', which omits the exponential and the alpha parameter shown in the RIE definition above it. Please correct the equation.
  5. [Fig. 5; Supp. C.4] The main text states that 'other metrics exhibit the same trend' for the one-hot versus predefined-feature comparison, but Supp. C.4 says the predefined features outperform 'in the majority of cases,' which is a weaker statement. Please align the main-text claim with the supplementary results.
  6. [Sec. 4.2; Supp. A.5] The choice of 1000 µM as a placeholder for inactive compounds in the three datasets with additional measurements is acknowledged in Supp. A.5, but the main text does not mention this artificial value or its potential effect on regression tasks. A one-sentence caveat in Sec. 4.1 or 4.2 would help readers who use the floating-value labels.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the curated dataset and benchmarking protocol are self-contained, and the cited same-group works are not load-bearing in any derivation.

full rationale

The paper's central contribution is a curated dataset and a standardized evaluation protocol. The curation pipeline is described algorithmically (duplicate removal, hierarchical primary/confirmatory/counter screening, PAINS and druglikeness filters, expert verification) and implemented in publicly available code, so the 'high-quality labels' claim is supported by an independent procedure rather than by definition. The benchmarking section compares existing models under a standard train/validation/test protocol; hyperparameters are tuned on validation folds and evaluated on held-out test sets, so no fitted parameter is renamed as a prediction. Self-citations are present but not load-bearing: same-group works [9,13] are cited for the general fact that primary HTS thresholds are loose and for bioassay identification, while the specific curation hierarchies, filters, and dataset statistics are produced by this paper's own pipeline; BCL [39] is used only as a comparison baseline. The 3D graph 6 Angstrom cutoff, logAUC range, and BEDROC alpha are acknowledged heuristics adopted from prior work, not results derived from the dataset. The paper explicitly acknowledges limitations such as the artificial 1000 µM inactive value (Supplement A.5), and Section 6 lists remaining limitations (imbalance, scaffold shift, conformations); these are data-quality caveats, not circular reasoning. The skeptic's concern that most inactives come from unconfirmed primary-screen negatives is a plausible correctness risk for the 'clean labels' premise, but it is an empirical claim about assay noise, not a tautology or a fit; it does not make the derivation circular. The paper's own retest examples in Supplement A.2 are treated as curation decisions rather than hidden circular constraints, and no claim in the paper reduces by construction to its inputs.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

The central benchmark claim does not rest on a fitted numerical model, so there are no fitted free parameters in the usual sense. The listed entries are hand-chosen protocol constants that define the evaluation. The main domain assumptions concern the reliability of PubChem metadata and the unvalidated treatment of primary-screen inactives as true inactives. No new physical entities are introduced.

free parameters (5)
  • Inactive placeholder IC50/EC50 = 1000 µM (assigned)
    For three datasets with continuous activity values, inactive compounds are assigned 1000 µM (Sec. A.5). It is a hand-set constant, not fitted, and does not affect the binary classification benchmark.
  • BEDROC alpha = 20
    Set to the value recommended by the metric's original paper (Sec. B.1). Controls how strongly early rankings are rewarded.
  • logAUC FPR range = [0.001, 0.1]
    Chosen following prior work (Sec. B.1). Defines the early-FPR region the metric integrates over.
  • 3D graph distance cutoff = 6 Å
    Used to define edges in 3D graphs, following prior work (Sec. 3.3). Affects graph connectivity and hence model input.
  • BM scaffold training assignment threshold = >10% of dataset in a scaffold bin
    Standardized rule for scaffold splits (Sec. 4.3). Any scaffold bin larger than 10% of total molecules is placed in training; ad hoc but explicit.
assumptions (5)
  • domain assumption PubChem bioassay records and their textual descriptions accurately encode the relationships among primary, confirmatory, and counter screens.
    Hierarchical curation relies on manual reading of PubChem assay descriptions (Sec. 3.2, supplement Sec. A.2). If these records are incomplete or misread, active/inactive labels inherit the error.
  • ad hoc to paper Primary-screen inactivity is treated as true inactivity in most final inactive sets.
    Most pipelines take final inactives from primary-screen inactive readouts only (e.g., AID1798 pipeline, Fig. 10). This is a practical but unvalidated assumption given known primary-screen false negatives.
  • domain assumption PAINS, promiscuity, and druglikeness filters remove artifacts without removing a meaningful number of real actives.
    These filters are applied globally (Sec. 3.2). The paper reports only Corina's filtered-out molecule count, not a detailed analysis of how many actives the filters removed.
  • domain assumption A low-energy 3D conformation generated by Corina is a sufficient structural representation for benchmarking 3D models.
    The standard 3D format assumes the molecule is at its lowest energy state (Sec. 6 acknowledges this and notes binding conformations as future work).
  • ad hoc to paper Adapted cross-validation with the validation fold fixed as the fold preceding the test fold gives estimates comparable to nested cross-validation.
    Sec. 4.3 and Sec. B.2 propose this for computational efficiency without an empirical check that it preserves nested-CV calibration.

how reviews work

0 comments
Cite this review

Pith. "Pith review of WelQrate: Defining the Gold Standard in Small Molecule Drug Discovery Benchmarking." pith.science (2026). https://pith.science/paper/WWW42MF5

@misc{pith2026241109820,
  author       = {Pith},
  title        = {Pith review of: WelQrate: Defining the Gold Standard in Small Molecule Drug Discovery Benchmarking},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WWW42MF5}},
  note         = {Machine review of arXiv:2411.09820}
}
read the original abstract

While deep learning has revolutionized computer-aided drug discovery, the AI community has predominantly focused on model innovation and placed less emphasis on establishing best benchmarking practices. We posit that without a sound model evaluation framework, the AI community's efforts cannot reach their full potential, thereby slowing the progress and transfer of innovation into real-world drug discovery. Thus, in this paper, we seek to establish a new gold standard for small molecule drug discovery benchmarking, WelQrate. Specifically, our contributions are threefold: WelQrate Dataset Collection - we introduce a meticulously curated collection of 9 datasets spanning 5 therapeutic target classes. Our hierarchical curation pipelines, designed by drug discovery experts, go beyond the primary high-throughput screen by leveraging additional confirmatory and counter screens along with rigorous domain-driven preprocessing, such as Pan-Assay Interference Compounds (PAINS) filtering, to ensure the high-quality data in the datasets; WelQrate Evaluation Framework - we propose a standardized model evaluation framework considering high-quality datasets, featurization, 3D conformation generation, evaluation metrics, and data splits, which provides a reliable benchmarking for drug discovery experts conducting real-world virtual screening; Benchmarking - we evaluate model performance through various research questions using the WelQrate dataset collection, exploring the effects of different models, dataset quality, featurization methods, and data splitting strategies on the results. In summary, we recommend adopting our proposed WelQrate as the gold standard in small molecule drug discovery benchmarking. The WelQrate dataset collection, along with the curation codes, and experimental scripts are all publicly available at WelQrate.org.

Figures

Figures reproduced from arXiv: 2411.09820 by the authors.

Figure 1
Figure 1. An overview of the data curation pipeline. Secondly, WelQrate dataset collection employs a rigorous curation pipeline to ensure high data quality [9]. The initial primary HTS has a high false positive rate. Therefore, in a real-world HTS campaign, a series of follow-up screens are carried out to ensure the correctness and relevance of the data. PubChem [10], a publicly accessible database of chemical molecules and t… view at source ↗
Figure 2
Figure 2. An example of the hierarchical curation with AID 1798. Initially 63,676 compounds go through a primary screen (AID 626). The found 1,665 actives fur￾ther go through a confirmatory screen (AID 1488) to verify their activities, and those showing activity in a counter screen (AID 1741) are excluded from the final active set. Additional data formats are provided for fair benchmark￾ing. However, researchers are encourage… view at source ↗
Figure 3
Figure 3. Illustration of the adapted cross-valiation. We propose two standard dataset split methods for benchmarking: random and scaffold. Given that dataset splits significantly impact model performance, we recommend nested cross-validation as the ideal standard for random splits if resources allow, as it ensures robust evaluation. However, recognizing computational constraints, we also suggest an alternative approach that … view at source ↗
Figures from the paper (16 more)
Figure 4
Figure 4. Figure 4: Categorical performance comparison among different models (RQ1) trained respectively with [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: Comparison of model performance using one-hot [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: Comparison of model performance under random and scaffold split (RQ4). Error bars denote [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]
Figure 7
Figure 7. Figure 7: An example of inconsistent molecular representations for the same molecule found in [PITH_FULL_IMAGE:figures/full_fig_p016_7.png]
Figure 8
Figure 8. Figure 8: An example of undefined stereochemistry MoleculeNet’s BACE dataset. The highlighted circle [PITH_FULL_IMAGE:figures/full_fig_p016_8.png]
Figure 9
Figure 9. Figure 9: Hierarchical curation pipeline for AID435008 [PITH_FULL_IMAGE:figures/full_fig_p018_9.png]
Figure 10
Figure 10. Figure 10: Hierarchical curation pipeline for AID1798 [PITH_FULL_IMAGE:figures/full_fig_p019_10.png]
Figure 11
Figure 11. Figure 11: Hierarchical curation pipeline for AID435034 [PITH_FULL_IMAGE:figures/full_fig_p020_11.png]
Figure 12
Figure 12. Figure 12: Hierarchical curation pipeline for AID1843 [PITH_FULL_IMAGE:figures/full_fig_p021_12.png]
Figure 13
Figure 13. Figure 13: Hierarchical curation pipeline for AID2258 [PITH_FULL_IMAGE:figures/full_fig_p022_13.png]
Figure 14
Figure 14. Figure 14: Hierarchical curation pipeline for AID463087 [PITH_FULL_IMAGE:figures/full_fig_p023_14.png]
Figure 15
Figure 15. Figure 15: Hierarchical curation pipeline for AID488997 [PITH_FULL_IMAGE:figures/full_fig_p024_15.png]
Figure 16
Figure 16. Figure 16: Hierarchical curation pipeline for AID2689 [PITH_FULL_IMAGE:figures/full_fig_p025_16.png]
Figure 17
Figure 17. Figure 17: Hierarchical curation pipeline for AID485290 [PITH_FULL_IMAGE:figures/full_fig_p026_17.png]
Figure 18
Figure 18. Figure 18: A T-SNE visualization of the ECPF4 embedding of AID1798, before and after curation. [PITH_FULL_IMAGE:figures/full_fig_p029_18.png]
Figure 19
Figure 19. Figure 19: Comparison of model performance using one-hot encoding and pre-defined features in [PITH_FULL_IMAGE:figures/full_fig_p038_19.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

58 extracted references · 51 canonical work pages

  1. [1]

    https://www.meddra.org/

    MEDDRA. https://www.meddra.org/. [Online; accessed 05-June-2024]

    work page 2024

  2. [2]

    Comprehensive behavioral analysis of male ox1r-/- mice showed implication of orexin receptor-1 in mood, anxiety, and social behavior

    Md G Abbas, Hirotaka Shoji, Shingo Soya, Mari Hondo, Tsuyoshi Miyakawa, and Takeshi Sakurai. Comprehensive behavioral analysis of male ox1r-/- mice showed implication of orexin receptor-1 in mood, anxiety, and social behavior. Frontiers in behavioral neuroscience, 9:324, 2015

    work page 2015

  3. [3]

    Smitha Antony, Christophe Marchand, Andrew G Stephen, Laurent Thibaut, Keli K Agama, Robert J Fisher, and Yves Pommier. Novel high-throughput electrochemiluminescent assay for identification of human tyrosyl-dna phosphodiesterase (tdp1) inhibitors and characterization of furamidine (nsc 305831) as an inhibitor of tdp1. Nucleic acids research, 35(13):4474–...

    work page 2007

  4. [4]

    Muscarinic antagonist control of myopia: evidence for m4 and m1 receptor-based pathways in the inhibition of experimentally-induced axial myopia in the tree shrew

    Baskar Arumugam and Neville A McBrien. Muscarinic antagonist control of myopia: evidence for m4 and m1 receptor-based pathways in the inhibition of experimentally-induced axial myopia in the tree shrew. Investigative Ophthalmology & Visual Science, 53(9):5827–5837, 2012

    work page 2012

  5. [5]

    New substructure filters for removal of pan assay interference compounds (pains) from screening libraries and for their exclusion in bioassays

    Jonathan B Baell and Georgina A Holloway. New substructure filters for removal of pan assay interference compounds (pains) from screening libraries and for their exclusion in bioassays. Journal of medicinal chemistry, 53(7):2719–2740, 2010

    work page 2010

  6. [6]

    The selective muscarinic agonist xanomeline improves both the cognitive deficits and behavioral symptoms of alzheimer disease

    NC Bodick, WW Offen, HE Shannon, J Satterwhite, R Lucas, R Van Lier, and SM Paul. The selective muscarinic agonist xanomeline improves both the cognitive deficits and behavioral symptoms of alzheimer disease. Alzheimer disease and associated disorders, 11:S16–22, 1997

    work page 1997

  7. [7]

    Chemical modulation of kv7 potassium channels

    Matteo Borgini, Pravat Mondal, Ruiting Liu, and Peter Wipf. Chemical modulation of kv7 potassium channels. RSC medicinal chemistry, 12(4):483–537, 2021

    work page 2021

  8. [8]

    The serine/threonine kinase stk33 exhibits autophosphorylation and phosphorylates the intermediate filament protein vimentin

    Bastienne Brauksiepe, Alejandro O Mujica, Harald Herrmann, and Erwin R Schmidt. The serine/threonine kinase stk33 exhibits autophosphorylation and phosphorylates the intermediate filament protein vimentin. BMC biochemistry, 9:1–12, 2008

    work page 2008

Show all 58 references
  1. [9]

    Molecular structures: perception, autocorrelation descriptor and sar studies

    P Broto, G Moreau, and C Vandycke. Molecular structures: perception, autocorrelation descriptor and sar studies. perception of molecules: topological structure and 3-dimensional structure. European journal of medicinal chemistry, 19(1):61–65, 1984

    1984

  2. [10]

    Muscarinic m1 receptor-stimulated adenylate cyclase activity in chinese hamster ovary cells is mediated by gs α and is not a consequence of phosphoinositidase c activation

    Neil T BURFORD and Stefan R NAHORSKI. Muscarinic m1 receptor-stimulated adenylate cyclase activity in chinese hamster ovary cells is mediated by gs α and is not a consequence of phosphoinositidase c activation. Biochemical Journal, 315(3):883–888, 1996

    1996

  3. [11]

    High-throughput screening assay datasets from the pubchem database

    Mariusz Butkiewicz, Yanli Wang, Stephen H Bryant, Edward W Lowe Jr, David C Weaver, and Jens Meiler. High-throughput screening assay datasets from the pubchem database. Chemical informatics (Wilmington, Del.), 3(1), 2017

    2017

  4. [12]

    Targeting t-type/cav3

    Song Cai, Kimberly Gomez, Aubin Moutal, and Rajesh Khanna. Targeting t-type/cav3. 2 channels for chronic pain. Translational Research, 234:20–30, 2021

    2021

  5. [13]

    3d structure generator corina classic

    Structure Generator CORINA Classic. 3d structure generator corina classic. Nürnberg: Molecular Networks GmbH. Available online at: www. mn-am. com, 2019

    2019

  6. [14]

    Biopython: freely available python tools for computational molecular biology and bioinformatics

    Peter JA Cock, Tiago Antao, Jeffrey T Chang, Brad A Chapman, Cymon J Cox, Andrew Dalke, Iddo Friedberg, Thomas Hamelryck, Frank Kauff, Bartek Wilczynski, et al. Biopython: freely available python tools for computational molecular biology and bioinformatics. Bioinformatics, 25(...

    2009

  7. [15]

    Unique kir2

    Amit S Dhamoon, Sandeep V Pandit, Farzad Sarmast, Keely R Parisian, Prabal Guha, You Li, Suveer Bagwe, Steven M Taffet, and Justus MB Anumonwo. Unique kir2. x properties determine regional and species differences in the cardiac inward rectifier k+ current. Circulation research...

    2004

  8. [16]

    Blockade of orexin-1 receptors atten- uates orexin-2 receptor antagonism-induced sleep promotion in the rat

    Christine Dugovic, Jonathan E Shelton, Leah E Aluisio, Ian C Fraser, Xiaohui Jiang, Steven W Sutton, Pascal Bonaventure, Sujin Yun, Xiaorong Li, Brian Lord, et al. Blockade of orexin-1 receptors atten- uates orexin-2 receptor antagonism-induced sleep promotion in the rat. Jour...

    2009

  9. [17]

    Pytorch geometric, 2019

    Matthias Fey and Jan Eric Lenssen. Pytorch geometric, 2019. URL https://github.com/rusty1s/ pytorch_geometric. Accessed: 2024-06-11

    2019

  10. [18]

    Deep learning for virtual screening: Five reasons to use roc cost functions

    Vladimir Golkov, Alexander Becker, Daniel T Plop, Daniel ˇCuturilo, Neda Davoudi, Jeffrey Mendenhall, Rocco Moretti, Jens Meiler, and Daniel Cremers. Deep learning for virtual screening: Five reasons to use roc cost functions. arXiv preprint arXiv:2007.07029, 2020. 40

    2007 arXiv

  11. [19]

    International union of pharmacology

    George A Gutman, K George Chandy, John P Adelman, Jayashree Aiyar, Douglas A Bayliss, David E Clapham, Manuel Covarriubias, Gary V Desir, Kiyoshi Furuichi, Barry Ganetzky, et al. International union of pharmacology. xli. compendium of voltage-gated ion channels: potassium chan...

    2003

  12. [20]

    Alzheimer’s disease: targeting the cholinergic system

    Talita H Ferreira-Vieira, Isabella M Guimaraes, Flavia R Silva, and Fabiola M Ribeiro. Alzheimer’s disease: targeting the cholinergic system. Current neuropharmacology, 14(1):101–115, 2016

    2016

  13. [21]

    Orexin a activates locus coeruleus cell firing and increases arousal in the rat

    Jim J Hagan, Ron A Leslie, Sara Patel, Martyn L Evans, Trevor A Wattam, Steve Holmes, Christopher D Benham, Stephen G Taylor, Carol Routledge, Panida Hemmati, et al. Orexin a activates locus coeruleus cell firing and increases arousal in the rat. Proceedings of the National Ac...

    1999

  14. [22]

    Glide: a new approach for rapid, accurate docking and scoring

    Thomas A Halgren, Robert B Murphy, Richard A Friesner, Hege S Beard, Leah L Frye, W Thomas Pollard, and Jay L Banks. Glide: a new approach for rapid, accurate docking and scoring. 2. enrichment factors in database screening. Journal of medicinal chemistry, 47(7):1750–1759, 2004

    2004

  15. [23]

    Genetic variation of cacna1h in idiopathic generalized epilepsy.Annals of neurology, 55(4):595–596, 2004

    Sarah E Heron, Hilary A Phillips, John C Mulley, Aziz Mazarib, Miriam Y Neufeld, Samuel F Berkovic, and Ingrid E Scheffer. Genetic variation of cacna1h in idiopathic generalized epilepsy.Annals of neurology, 55(4):595–596, 2004

    2004

  16. [24]

    Therapeutics data commons: Machine learning datasets and tasks for drug discovery and development

    Kexin Huang, Tianfan Fu, Wenhao Gao, Yue Zhao, Yusuf Roohani, Jure Leskovec, Connor W Coley, Cao Xiao, Jimeng Sun, and Marinka Zitnik. Therapeutics data commons: Machine learning datasets and tasks for drug discovery and development. arXiv preprint arXiv:2102.09548, 2021

    2021 arXiv

  17. [25]

    Familial kcnq2 mutation: a psychiatric perspective

    Anton Iftimovici, Angeline Charmet, Béatrice Desnous, Ana Ory, Richard Delorme, Charles Coutton, Françoise Devillard, Mathieu Milh, and Anna Maruani. Familial kcnq2 mutation: a psychiatric perspective. Psychiatric Genetics, 34(1):24–27, 2024

    2024

  18. [26]

    Regulation of neuronal t-type calcium channels

    Mircea C Iftinca and Gerald W Zamponi. Regulation of neuronal t-type calcium channels. Trends in pharmacological sciences, 30(1):32–40, 2009

    2009

  19. [27]

    Ir evaluation methods for retrieving highly relevant documents

    Kalervo Järvelin and Jaana Kekäläinen. Ir evaluation methods for retrieving highly relevant documents. In ACM SIGIR Forum, volume 51, pages 243–250. ACM New York, NY , USA, 2017

    2017

  20. [28]

    Identification and characterization of the rat m1 muscarinic receptor promoter

    Christoph PR Klett and Tom I Bonner. Identification and characterization of the rat m1 muscarinic receptor promoter. Journal of neurochemistry, 72(3):900–909, 1999

    1999

  21. [29]

    The sider database of drugs and side effects

    Michael Kuhn, Ivica Letunic, Lars Juhl Jensen, and Peer Bork. The sider database of drugs and side effects. Nucleic acids research, 44(D1):D1075–D1079, 2016

    2016

  22. [30]

    Rdkit documentation

    Greg Landrum. Rdkit documentation. Release, 1(1-79):4, 2013

    2013

  23. [31]

    The embl-ebi bioinformatics web and programmatic tools framework

    Weizhong Li, Andrew Cowley, Mahmut Uludag, Tamer Gur, Hamish McWilliam, Silvano Squizzato, Young Mi Park, Nicola Buso, and Rodrigo Lopez. The embl-ebi bioinformatics web and programmatic tools framework. Nucleic acids research, 43(W1):W580–W584, 2015

    2015

  24. [32]

    Inhibition of human tyrosyl-dna phosphodiesterase by aminoglycoside antibiotics and ribosome inhibitors

    Zhiyong Liao, Laurent Thibaut, Andrew Jobson, and Yves Pommier. Inhibition of human tyrosyl-dna phosphodiesterase by aminoglycoside antibiotics and ribosome inhibitors. Molecular pharmacology, 70 (1):366–372, 2006

    2006

  25. [33]

    Spherical message passing for 3d graph networks

    Yi Liu, Limei Wang, Meng Liu, Xuan Zhang, Bora Oztekin, and Shuiwang Ji. Spherical message passing for 3d graph networks. arXiv preprint arXiv:2102.05013, 2021

    2021 arXiv

  26. [34]

    Interpretable chirality-aware graph neural network for quantitative structure activity relationship modeling in drug discovery

    Yunchao Lance Liu, Yu Wang, Oanh Vu, Rocco Moretti, Bobby Bodenheimer, Jens Meiler, and Tyler Derr. Interpretable chirality-aware graph neural network for quantitative structure activity relationship modeling in drug discovery. In Proceedings of the AAAI Conference on Artifici...

    2023

  27. [35]

    Effects of central muscarinic-1 receptor stimulation on blood pressure regulation

    Aharon Medina, Neil Bodick, Ary L Goldberger, Margaret Mac Mahon, and Lewis A Lipsitz. Effects of central muscarinic-1 receptor stimulation on blood pressure regulation. Hypertension, 29(3):828–834, 1997

    1997

  28. [36]

    John J Millichap and Edward C Cooper. Kcnq2 potassium channel epileptic encephalopathy syndrome: divorce of an electro-mechanical couple? kcnq2 potassium channel epileptic encephalopathy syndrome: divorce of an electro-mechanical couple? Epilepsy Currents, 12(4):150–152, 2012

    2012

  29. [37]

    The autocorrelation of a topological structure: A new molecular descriptor

    Gilles Moreau and Pierre Broto. The autocorrelation of a topological structure: A new molecular descriptor. 1980. 41

    1980

  30. [38]

    Rapid context-dependent ligand desolvation in molecular docking

    Michael M Mysinger and Brian K Shoichet. Rapid context-dependent ligand desolvation in molecular docking. Journal of chemical information and modeling, 50(9):1561–1573, 2010

    2010

  31. [39]

    The role of t-type calcium channels in epilepsy and pain

    MT Nelson, SM Todorovic, and E Perez-Reyes. The role of t-type calcium channels in epilepsy and pain. Current pharmaceutical design, 12(18):2189–2197, 2006

    2006

  32. [40]

    High-affinity choline transporter

    Takashi Okuda and Tatsuya Haga. High-affinity choline transporter. Neurochemical research, 28:483–488, 2003

    2003

  33. [41]

    Improved scoring of ligand- protein interactions using owfeg free energy grids

    David A Pearlman and Paul S Charifson. Improved scoring of ligand- protein interactions using owfeg free energy grids. Journal of medicinal chemistry, 44(4):502–511, 2001

    2001

  34. [42]

    Molecular physiology of low-voltage-activated t-type calcium channels

    Edward Perez-Reyes. Molecular physiology of low-voltage-activated t-type calcium channels. Physiologi- cal reviews, 83(1):117–161, 2003

    2003

  35. [43]

    Acetylcholine as a neuromodulator: cholinergic signaling shapes nervous system function and behavior

    Marina R Picciotto, Michael J Higley, and Yann S Mineur. Acetylcholine as a neuromodulator: cholinergic signaling shapes nervous system function and behavior. Neuron, 76(1):116–129, 2012

    2012

  36. [44]

    Massively multitask networks for drug discovery

    Bharath Ramsundar, Steven Kearnes, Patrick Riley, Dale Webster, David Konerding, and Vijay Pande. Massively multitask networks for drug discovery. arXiv preprint arXiv:1502.02072, 2015

    2015 arXiv

  37. [45]

    Maximum unbiased validation (muv) data sets for virtual screening based on pubchem bioactivity data

    Sebastian G Rohrer and Knut Baumann. Maximum unbiased validation (muv) data sets for virtual screening based on pubchem bioactivity data. Journal of chemical information and modeling, 49(2):169–184, 2009

    2009

  38. [46]

    Synthetic lethal interaction between oncogenic kras dependency and stk33 suppression in human cancer cells

    Claudia Scholl, Stefan Fröhling, Ian F Dunn, Anna C Schinzel, David A Barbie, So Young Kim, Serena J Silver, Pablo Tamayo, Raymond C Wadlow, Sridhar Ramaswamy, et al. Synthetic lethal interaction between oncogenic kras dependency and stk33 suppression in human cancer cells. Ce...

    2009

  39. [47]

    Fast, scalable generation of high-quality protein multiple sequence alignments using clustal omega

    Fabian Sievers, Andreas Wilm, David Dineen, Toby J Gibson, Kevin Karplus, Weizhong Li, Rodrigo Lopez, Hamish McWilliam, Michael Remmert, Johannes Söding, et al. Fast, scalable generation of high-quality protein multiple sequence alignments using clustal omega. Molecular system...

    2011

  40. [48]

    Fluorescence spectroscopic profiling of compound libraries

    Anton Simeonov, Ajit Jadhav, Craig J Thomas, Yuhong Wang, Ruili Huang, Noel T Southall, Paul Shinn, Jeremy Smith, Christopher P Austin, Douglas S Auld, et al. Fluorescence spectroscopic profiling of compound libraries. Journal of medicinal chemistry, 51(8):2363–2371, 2008

    2008

  41. [49]

    Orexin/hypocretin signaling at the orexin 1 receptor regulates cue-elicited cocaine-seeking

    Rachel J Smith, Ronald E See, and Gary Aston-Jones. Orexin/hypocretin signaling at the orexin 1 receptor regulates cue-elicited cocaine-seeking. European Journal of Neuroscience, 30(3):493–503, 2009

    2009

  42. [50]

    Chronic inhibition of cardiac kir2

    Haiyan Sun, Xiaodong Liu, Qiaojie Xiong, Sojin Shikano, and Min Li. Chronic inhibition of cardiac kir2. 1 and herg potassium channels by celastrol with dual effects on both ion conductivity and protein trafficking. Journal of biological chemistry, 281(9):5877–5884, 2006

    2006

  43. [51]

    We Need Better Benchmarks for Machine Learning in Drug Discovery

    Pat Walters. We Need Better Benchmarks for Machine Learning in Drug Discovery. https://practicalcheminformatics.blogspot.com/2023/08/ we-need-better-benchmarks-for-machine.html?m=1 , 2023. [Online; accessed 05-June-2024]

    2023

  44. [52]

    Deep graph library (dgl), 2019

    Minjie Wang, Jinjing Zhou, Quan Gan, and Zheng Zhang. Deep graph library (dgl), 2019. URL https://www.dgl.ai. Accessed: 2024-06-11

    2019

  45. [53]

    Database resources of the national center for biotechnology information

    David L Wheeler, Tanya Barrett, Dennis A Benson, Stephen H Bryant, Kathi Canese, Vyacheslav Chetvernin, Deanna M Church, Michael DiCuccio, Ron Edgar, Scott Federhen, et al. Database resources of the national center for biotechnology information. Nucleic acids research, 36(supp...

    2007

  46. [54]

    Alzheimer disease: evidence for selective loss of cholinergic neurons in the nucleus basalis

    Peter J Whitehouse, Donald L Price, Arthur W Clark, Joseph T Coyle, and Mahlon R DeLong. Alzheimer disease: evidence for selective loss of cholinergic neurons in the nucleus basalis. Annals of Neurology: Official Journal of the American Neurological Association and the Child N...

    1981

  47. [55]

    Differential effects of m1 and m2 receptor antagonists in perirhinal cortex on visual recognition memory in monkeys

    Wei Wu, Richard C Saunders, Mortimer Mishkin, and Janita Turchi. Differential effects of m1 and m2 receptor antagonists in perirhinal cortex on visual recognition memory in monkeys. Neurobiology of learning and memory, 98(1):41–46, 2012

    2012

  48. [56]

    Moleculenet: a benchmark for molecular machine learning

    Zhenqin Wu, Bharath Ramsundar, Evan N Feinberg, Joseph Gomes, Caleb Geniesse, Aneesh S Pappu, Karl Leswing, and Vijay Pande. Moleculenet: a benchmark for molecular machine learning. Chemical science, 9(2):513–530, 2018

    2018

  49. [57]

    Frequent hitters: nuisance artifacts in high-throughput screening

    Zi-Yi Yang, Jun-Hong He, Ai-Ping Lu, Ting-Jun Hou, and Dong-Sheng Cao. Frequent hitters: nuisance artifacts in high-throughput screening. Drug discovery today, 25(4):657–667, 2020

    2020

  50. [58]

    Novel neuroprotective k+ channel inhibitor identified by high-throughput screening in yeast

    Elena Zaks-Makhina, Yonjung Kim, Elias Aizenman, and Edwin S Levitan. Novel neuroprotective k+ channel inhibitor identified by high-throughput screening in yeast. Molecular pharmacology, 65(1): 214–219, 2004. 42

    2004

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.