Similarity-Quantized Relative Difference Learning for Improved Molecular Activity Prediction

Bharath Sathappan; Colin A. Grambow; Ewan Wallace; Kangway V. Chuang; Karina Zadorozhny; Vishnu Sresht

REVIEW 3 major objections 5 minor 53 references

Similarity-Quantized Relative Difference Learning for Improved Molecular Activity Prediction

T0 review · 3 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read Similarity-thresholded relative difference learning lifts neural molecular activity prediction across every tested architecture.

desk verdict SQRL is a solid empirical extension of pairwise difference learning, with consistent Spearman gains for deep models on activity benchmarks, but the underspecified α selection and the inference rule that applies the delta model to neighbors beyond the training threshold need closer scrutiny before the low-data generalization claim is taken at face value. read the letter →

arxiv 2501.09103 v1 pith:UA7O3RBY submitted 2025-01-15 cs.LG

Karina Zadorozhny , Kangway V. Chuang , Bharath Sathappan , Ewan Wallace , Vishnu Sresht , Colin A. Grambow This is my paper · ORCID

classification cs.LG

keywords molecularactivitypredictionrelativedifferencelearningsimilaritythresholdgraphneuralnetworkscliffslow-datadrugdiscoverypairwiseTanimoto

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 claims that molecular activity prediction improves when a model is trained to predict the difference in potency between pairs of similar molecules, rather than the absolute potency of each molecule. It introduces SQRL, a framework that builds training pairs from molecules whose structural distance falls below a threshold α, trains a network on representation differences, and at inference anchors a new molecule's prediction to its nearest training neighbor's observed activity. On 30 MoleculeACE activity tasks, the activity-cliff subset, and five proprietary drug discovery targets, every tested deep learning architecture showed consistent Spearman rank-correlation gains, with the largest jumps for models that start weak, for example Uni-Mol rising from 0.26 to 0.69. The paper argues this works because structurally similar pairs carry the most informative local signal in low-data, noisy settings.

What carries the argument

The load-bearing mechanism is the similarity-quantized relative representation: a paired dataset restricted by d(xi, xj) ≤ α and a model that consumes the representation difference g(xi) − g(xj). The threshold α controls a quantity-versus-relevance trade-off; the paper proposes selecting it from the training distance distribution, choosing a value smaller than the average pairwise distance, and shows that MAE improves as α grows to a point and then degrades as more dissimilar pairs are added. At inference the same distance function selects the anchor molecule, converting a global regression problem into a local, anchored one.

What would settle it

Split SQRL's test predictions by the distance from each test molecule to its nearest training neighbor; if the reported Spearman gains over standard training disappear or reverse for test molecules whose anchor lies beyond the training threshold α, the anchor-extrapolation assumption fails. A concrete experiment would rerun Table 1 while excluding such test molecules and check whether the gains vanish.

Watch

Extended reading notes

Core claim

The paper's central discovery is that reformulating molecular activity prediction as similarity-thresholded relative difference learning converts weak neural predictors into competitive ones. Given a distance function d and threshold α, SQRL trains on the paired dataset Drel = {((xi, xj), yi − yj) | d(xi, xj) ≤ α}, minimizing mean squared error over f(g(xi) − g(xj)). At test time a new molecule xnew is assigned the prediction y_new = (1/n) Σ_{xi ∈ NNn(xnew)} (yi + f(g(xi) − g(xnew))). The empirical result is that this objective improves ranking accuracy and mean absolute error for MLPs, graph neural networks, and transformer-based models across public and internal activity datasets, while tree-based and nearest-neighbor baselines are not helped. The improvement is largest on activity-cliff molecules, where structurally similar compounds have very different activities, indicating that the method captures local structure-activity relationships that absolute regression objectives miss.

Load-bearing premise

The test-time prediction assumes that the nearest training neighbor's measured activity is a reliable anchor and that the model trained on pairs within distance α can be extrapolated to that neighbor even when the neighbor lies farther than α.

Editorial extensions

If this is right

Neural networks trained with SQRL show consistent Spearman rank-correlation gains across all tested architectures and datasets, with the weakest models gaining the most: Uni-Mol from 0.26 to 0.69, GINE from 0.33 to 0.76, MolCLR from 0.35 to 0.77, and MLP from 0.32 to 0.73 on MoleculeACE.
SQRL produces the largest gains for test molecules close to training data, while roughly maintaining standard performance for more distant points, as shown in the distance-conditioned analysis.
The improvements transfer to proprietary internal drug discovery targets, not just public benchmarks, supporting real-world applicability.
Tree-based baselines such as XGBoost, random forest, and k-nearest neighbors do not benefit from SQRL, which the paper attributes to the simple difference-fingerprint representation discarding the rest of the molecular structure.
Distance metrics with left-skewed or high-kurtosis pairwise distance distributions, such as Tanimoto distance on Morgan fingerprints and COATI embeddings, give the best results under SQRL.

Reading between the lines

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

SQRL's inference-time anchor, the nearest training neighbor, is used even when that neighbor lies beyond the training threshold α; a natural extension is to condition predictions on anchor distance or to abstain when no sufficiently close anchor exists.
The method's architecture-agnostic formulation suggests it could transfer to other low-data molecular endpoints, such as ADMET or solubility prediction, with any embedding-based similarity metric.
Because tree models are not helped by the difference representation, concatenating full molecular representations with difference representations may extend SQRL's gains to non-neural models, a possibility the paper hints at but does not test.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

The real news here is simple: similarity-thresholded relative difference learning helps deep models on molecular activity prediction, and the help is consistent across architectures and datasets. Table 1 is the paper's backbone, and it holds up — every GNN and transformer gains in Spearman rank correlation on MoleculeACE, MoleculeACE-Cliff, and the internal targets, with the biggest jumps on activity cliffs. Figure 6 gives some mechanism: the gains concentrate on test points with close training neighbors, which is what you'd expect if learning local deltas is the operative effect. That is a genuinely new empirical result, going beyond the indiscriminate pairwise training in Tynes, Wetzel, and Fralish. The paper also cites that prior work fairly and is honest about the assumption that good distance metrics exist. There's no circularity: the pairwise objective is not the same as the ranking evaluation, so the gains are not forced by construction.

The soft spots are real but addressable. The threshold α = 0.7 appears to be selected from the same sweep shown in Figure 1, and I don't see a nested validation that protects against overfitting to the test set. That's a common problem in this literature, and it needs stating. The bigger issue is the inference rule in Eq. 3: it always uses the nearest training neighbor, even when that neighbor is farther than α. The delta model never saw such large representation differences during training, and the prediction directly inherits the noise of the anchor label. Figure 6 partially covers this by showing Spearman as a function of nearest-neighbor distance, but it pools across tasks and doesn't report how many test points fall beyond α or how noisy the anchors are per task. Without that, the aggregate gains could be driven mainly by the oracle-like anchor for locally dense test molecules, with the learned delta contributing less than the headline implies. Those are fixable with a few extra experiments, not fatal.

There's also no code or data release and no seed-level variance, only standard deviations across tasks. That makes the numbers harder to trust than they should be.

Bottom line: this is a useful paper for anyone working on molecular activity prediction, especially low-data regimes and activity cliffs. It deserves a serious peer review, but I'd want the authors to show α selection via nested validation, report results conditioned on anchor distance, and release artifacts. Not a paradigm shift, but a solid, citable step forward.

Referee Report

3 major / 5 minor

Summary. The paper introduces Similarity-Quantized Relative Learning (SQRL), a framework in which molecular activity prediction is reformulated as learning relative activity differences between structurally similar molecule pairs. Training pairs are restricted by a distance threshold α on a molecular similarity metric, and test predictions are formed by adding the predicted difference to the activity of the nearest training molecule. The authors benchmark SQRL against standard absolute-value training across 30 MoleculeACE tasks, the MoleculeACE-Cliff subset, and five proprietary internal targets, using tree-based baselines, MLPs, GNNs, and transformer-based models. The central empirical claim is that SQRL consistently and substantially improves Spearman rank correlation for all deep learning architectures tested, with the largest gains for models that perform poorly under standard training, while not helping tree-based baselines. The paper also provides a mechanism analysis showing that SQRL's improvements concentrate on test points with nearby training neighbors.

Significance. If the results hold, the paper makes a useful empirical contribution: a simple, architecture-agnostic training reformulation that consistently improves ranking performance for neural molecular property predictors, with particular relevance to activity-cliff tasks and low-data regimes. The strengths of the paper are its breadth (multiple public benchmarks plus proprietary data), the inclusion of negative results for tree-based models, and the mechanistic evidence in Figure 6 that gains are localized to test points with close training neighbors. The framework is not circular in the sense that the pairwise training objective differs from the absolute-value evaluation metric, and the main gains are not forced by construction. However, the central claim rests on two load-bearing assumptions that are not fully validated: that α=0.7 is selected without test-set leakage, and that the nearest-neighbor anchoring step in Eq. (3) remains valid when the nearest neighbor lies outside the training-pair distance threshold α. These issues require additional analysis before the reported gains can be taken at face value.

major comments (3)

[Section 3, Eq. (3)] The inference rule in Eq. (3) uses the nearest training molecule with no restriction on its distance to the test molecule, while the training distribution in Eq. (1) contains only pairs with d(xi,xj) ≤ α. For any test molecule whose nearest training neighbor is farther than α, the model f is asked to extrapolate on a representation difference that never appeared in training, and the prediction directly inherits the noise of the anchor label y_NN. Figure 6 partially addresses this by showing Spearman correlation as a function of nearest-neighbor distance, but it pools across tasks and does not report how many test molecules fall beyond α, nor the anchor-label noise level per task. This is the most direct threat to the paper's 'low-data generalization' claim. The authors should report, per task and in aggregate, the fraction of test points with nearest-neighbor distance > α, and should condition the Table 1 results on whether the nearest neighbor lies within α. If the gains persist only for the within-α subset, the claim that SQRL improves general ranking should be substantially qualified.
[Section 4.2 and Table 1] The threshold α=0.7 used for all main results appears to be selected from the same threshold sweeps shown in Figure 1 and Appendix Figure 5, without any described nested validation or an explicit selection rule. Figure 1 shows MAE as a function of α for MLP models, and Table 1 then reports results at α=0.7; if α was chosen after inspecting test-set performance across thresholds, the reported gains are optimistically biased. The paper states a heuristic (choose α smaller than the average pairwise distance) but does not operationalize it into a reproducible selection procedure. The authors should either provide a nested validation protocol that selects α on held-out tasks or validation folds, or report results for a range of α values and show that the conclusions are robust.
[Appendix A.2, Table 2] The standard and SQRL variants differ not only in training objective but also in hyperparameters, hidden sizes, number of layers, dropout, learning rate, and batch size. For example, the standard MLP uses [256,256] with learning rate 1e-4, while MLP-SQRL uses [512,256] with learning rate 1e-5; similar differences appear for GINE, PNA, and COATI. The paper does not establish that the standard baselines were tuned with comparable effort; the reported Uni-Mol standard Spearman of 0.26 is far below typical published performance for stronger configurations, which raises the concern that the headline gains partly reflect undertrained baselines rather than a benefit of the SQRL objective. The authors should report hyperparameter search budgets for both settings, and ideally include a stronger standard baseline (e.g., fine-tuned Uni-Mol or a published reference value) to confirm that SQRL improves over reasonably tuned standard training.

minor comments (5)

[Appendix A.1] There is a typo in 'Morgan fingeprints' (missing 'r'), and the phrase 'Due to the computationally expensive conformation generation, Uni-Mol-SQRL was evaluated on the following subset' is duplicated across two places in the appendix.
[Section 3, Eq. (1)-(2)] The symbol d is overloaded: it denotes both the input-space distance function in Eq. (1) and the dimensionality of the representation space in the definition of g : X → R^d in Section 3. This makes the notation confusing; one of the two should be renamed.
[Figure 1] The top panel of Figure 1 plots MAE but the y-axis label appears incomplete ('MAE ( )'), and the caption does not state which model and features were used for the sweep; Appendix A.5 clarifies that this is an MLP, but the main text should say so.
[Appendix A.1, PNA] The text says the PNA implementation follows Ref. 28 (FS-Mol), but Ref. 28 is a few-shot learning dataset paper and is unlikely to be the correct source for the PNA implementation; the citation should be checked and corrected.
[General] The paper provides no code repository, data availability statement, or detailed training configuration (e.g., optimizer, early stopping, number of seeds) beyond the hyperparameter table; given the empirical nature of the contribution, sharing code or at least a detailed configuration file would materially strengthen reproducibility.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the pairwise training objective, anchored inference rule, and similarity threshold are distinct from the evaluation targets and are not fitted parameters renamed as predictions.

full rationale

The paper's derivation chain is self-contained. Training minimizes Eq. 2 over Drel pairs with d(xi, xj) ≤ α, while inference applies Eq. 3, anchoring at the observed label of the nearest training molecule. Using training labels as anchors is a deliberate part of the method, not a circular re-derivation of the test labels; the approach is benchmarked against standard absolute models and KNN with the same training data. The only apparent self-citation, Ref. [13] (a general representation-learning review co-authored by one of the present authors), is background context and does not carry the SQRL claim. No parameter is fitted to a target subset and then reported as a prediction: α is a threshold selected from training-distance statistics, and the main comparisons use a fixed α = 0.7 across tasks. The stated limitation that meaningful distance metrics must be available is an assumption, not a circular step. Open methodological concerns, such as out-of-threshold extrapolation in Eq. 3, anchor-label noise, and hyperparameter selection, are correctness risks rather than instances of definitional or self-citational circularity.

Assumptions & free parameters 3 free parameters · 3 assumptions · 0 invented entities

The central claim rests on one main free parameter, the distance threshold, plus standard model hyperparameters. The key domain assumption is that structural similarity proxies for the informativeness of a comparison; if that fails, the whole method loses its basis. No new physical entities are introduced.

free parameters (3)

Distance threshold alpha = 0.7 (Tanimoto) for main results
Controls which molecule pairs enter Drel. The paper proposes a training-distribution heuristic but the main tables use a fixed alpha=0.7, apparently chosen from the MAE sweep in Figure 1; no per-task nested validation is described.
Number of neighbors n at inference = 1 (default)
Number of training molecules used to anchor the prediction in Eq. 3. The paper states 'unless otherwise specified, we set n=1' but does not study sensitivity.
Model hyperparameters = per model, Table 2
Learning rates, layer sizes, dropout, batch size differ between Standard and SQRL variants; the optimization procedure is not described, so these are hand-chosen or searched parameters.

assumptions (3)

domain assumption Chemical similarity, as measured by Tanimoto distance on Morgan fingerprints (or other metrics), is a meaningful proxy for the informativeness of a training pair.
Section 3 builds Drel by thresholding pairwise distances, and Section 4.2 motivates the choice of metrics; the entire method assumes local pairs carry the most signal.
domain assumption The nearest training molecule's observed activity y_i is a reliable anchor for predicting y_new, and the delta model generalizes to pairs whose distance may exceed the training threshold alpha.
Eq. 3 evaluates on the nearest neighbor without restricting distance; Figure 6 partially tests this, showing SQRL is comparable to standard models for distant points.
ad hoc to paper The heuristic that alpha should be smaller than the average pairwise distance is sufficient for good performance.
Section 3: 'We propose selecting alpha based on the distribution of distances in the training data, specifically by choosing a threshold smaller than the average pairwise distance.' This heuristic is not validated per task.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Similarity-Quantized Relative Difference Learning for Improved Molecular Activity Prediction." pith.science (2026). https://pith.science/paper/UA7O3RBY

@misc{pith2026250109103,
  author       = {Pith},
  title        = {Pith review of: Similarity-Quantized Relative Difference Learning for Improved Molecular Activity Prediction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UA7O3RBY}},
  note         = {Machine review of arXiv:2501.09103}
}

read the original abstract

Accurate prediction of molecular activities is crucial for efficient drug discovery, yet remains challenging due to limited and noisy datasets. We introduce Similarity-Quantized Relative Learning (SQRL), a learning framework that reformulates molecular activity prediction as relative difference learning between structurally similar pairs of compounds. SQRL uses precomputed molecular similarities to enhance training of graph neural networks and other architectures, and significantly improves accuracy and generalization in low-data regimes common in drug discovery. We demonstrate its broad applicability and real-world potential through benchmarking on public datasets as well as proprietary industry data. Our findings demonstrate that leveraging similarity-aware relative differences provides an effective paradigm for molecular activity prediction.

Figures

Figures reproduced from arXiv: 2501.09103 by the authors.

**Figure 1.** Leveraging local structural information enhances predictive performance. Top: Incorporating neighbors only up to a certain distance threshold α improves MAE (↓). Bottom: Pairwise distance distributions of training data (overlaid for all 30 MoleculeACE tasks) with greater skewness and kurtosis yield the best performance and a wider range of acceptable values of α. Relative representation. We define g : X → R d as a m… view at source ↗

**Figure 2.** Training data sizes for each task in MoleculeACE. [PITH_FULL_IMAGE:figures/full_fig_p010_2.png] view at source ↗

**Figure 3.** Molecular pairs obtained by the data matching procedure described in Section 3 at different Tanimoto distance thresholds α for MoleculeACE task CHEMBL1862_Ki [PITH_FULL_IMAGE:figures/full_fig_p011_3.png] view at source ↗

Figures from the paper (2 more)

**Figure 5.** Figure 5: Leveraging local structural information enhances predictive performance. MAE (↓) as a function of distance threshold α for several additional distance metrics compared to [PITH_FULL_IMAGE:figures/full_fig_p012_5.png]

**Figure 6.** Figure 6: SQRL enhances local molecular consistency. Spearman’s rank correlation coefficient (↑) plotted as a function of the distance between test points and their nearest neighbors in the training set. SQRL-trained models that benefit from this training strategy demonstrate th…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

53 extracted references · 32 canonical work pages

[1]

Geometric deep learning on molecular repre- sentations

Kenneth Atz, Francesca Grisoni, and Gisbert Schneider. Geometric deep learning on molecular repre- sentations. Nature Machine Intelligence, 3(12):1023–1032, 2021. URL https://doi.org/10.1038/ s42256-021-00418-8

work page 2021
[2]

Schoenholz, Patrick F

Justin Gilmer, Samuel S. Schoenholz, Patrick F. Riley, Oriol Vinyals, and George E. Dahl. Neural message passing for quantum chemistry. In Proceedings of the 34th International Conference on Machine Learning - V olume 70, ICML’17, pages 1263–1272, 2017. URL https://proceedings.mlr.press/ v70/gilmer17a/gilmer17a.pdf

work page 2017
[3]

Analyzing learned molecular representations for property prediction

Kevin Yang, Kyle Swanson, Wengong Jin, Connor Coley, Philipp Eiden, Hua Gao, Angel Guzman-Perez, Timothy Hopper, Brian Kelley, Miriam Mathea, Andrew Palmer, V olker Settels, Tommi Jaakkola, Klavs Jensen, and Regina Barzilay. Analyzing learned molecular representations for property prediction. Journal of Chemical Information and Modeling , 59(8):3370–3388,...

work page 2019
[4]

Pushing the boundaries of molecular representation for drug discovery with the graph attention mechanism

Zhaoping Xiong, Dingyan Wang, Xiaohong Liu, Feisheng Zhong, Xiaozhe Wan, Xutong Li, Zhaojun Li, Xiaomin Luo, Kaixian Chen, Hualiang Jiang, and Mingyue Zheng. Pushing the boundaries of molecular representation for drug discovery with the graph attention mechanism. Journal of Medicinal Chemistry , 63 (16):8749–8760, 2020. URL https://doi.org/10.1021/acs.jme...

doi:10.1021/acs.jmedchem.9b00959 2020
[5]

Spherical message passing for 3D molecular graphs

Yi Liu, Limei Wang, Meng Liu, Yuchao Lin, Xuan Zhang, Bora Oztekin, and Shuiwang Ji. Spherical message passing for 3D molecular graphs. In International Conference on Learning Representations, 2022. URL https://openreview.net/forum?id=givsRXsOt9r

work page 2022
[6]

K. T. Schütt, P.-J. Kindermans, H. E. Sauceda, S. Chmiela, A. Tkatchenko, and K.-R. Müller. SchNet: A continuous-filter convolutional neural network for modeling quantum interactions. In Proceedings of the 31st International Conference on Neural Information Processing Systems , NIPS’17, pages 992–1002, Red Hook, NY , USA, 2017. Curran Associates Inc. URLh...

work page 2017
[7]

E(n) equivariant graph neural networks,

Victor Garcia Satorras, Emiel Hoogeboom, and Max Welling. E(n) equivariant graph neural networks,
[8]

SMILES-BERT: Large scale unsupervised pre-training for molecular property prediction

Sheng Wang, Yuzhi Guo, Yuhong Wang, Hongmao Sun, and Junzhou Huang. SMILES-BERT: Large scale unsupervised pre-training for molecular property prediction. In Proceedings of the 10th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics , BCB ’19, pages 429–436, New York, NY , USA, 2019. Association for Computing Machi...

arXiv 2019

Show all 53 references

[9]

Shion Honda, Shoi Shi, and Hiroki R. Ueda. SMILES Transformer: Pre-trained molecular fingerprint for low data drug discovery, 2019. URL https://arxiv.org/abs/1911.04738. 5

2019 arXiv
[10]

RoBERTa: A robustly optimized BERT pretraining approach,

Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, and Veselin Stoyanov. RoBERTa: A robustly optimized BERT pretraining approach,
[11]

ChemBERTa: Large-scale self-supervised pretraining for molecular property prediction, 2020

Seyone Chithrananda, Gabriel Grand, and Bharath Ramsundar. ChemBERTa: Large-scale self-supervised pretraining for molecular property prediction, 2020. URL https://arxiv.org/abs/2010.09885

2020 arXiv
[12]

Large-scale chemical language representations capture molecular structure and properties

Jerret Ross, Brian Belgodere, Vijil Chenthamarakshan, Inkit Padhi, Youssef Mroueh, and Payel Das. Large-scale chemical language representations capture molecular structure and properties. Nature Machine Intelligence, 4(12):1256–1264, 2022. URL https://doi.org/10.1038/s42256-02...

2022 doi
[13]

Chuang, Laura M

Kangway V . Chuang, Laura M. Gunsalus, and Michael J. Keiser. Learning molecular representations for medicinal chemistry. Journal of Medicinal Chemistry , 63(16):8705–8722, 2020. URL https: //doi.org/10.1021/acs.jmedchem.0c00385

2020 doi
[14]

Fundamentals of Medicinal Chemistry

Gareth Thomas. Fundamentals of Medicinal Chemistry . John Wiley & Sons, Chichester, England, 1st edition, March 2004

2004
[15]

Matched molecular pair analysis in drug discovery: Methods and recent applications

Ziyi Yang, Shaohua Shi, Li Fu, Aiping Lu, Tingjun Hou, and Dongsheng Cao. Matched molecular pair analysis in drug discovery: Methods and recent applications. Journal of Medicinal Chemistry , 66(7): 4361–4377, 2023. URL https://doi.org/10.1021/acs.jmedchem.2c01787

2023 doi
[16]

A systematic study of key elements underlying molecular property prediction

Jianyuan Deng, Zhibo Yang, Hehe Wang, Iwao Ojima, Dimitris Samaras, and Fusheng Wang. A systematic study of key elements underlying molecular property prediction. Nature Communications, 14(1), 2023. URL https://doi.org/10.1038/s41467-023-41948-6

2023 doi
[17]

Jun Xia, Lecheng Zhang, Xiao Zhu, Yue Liu, Zhangyang Gao, Bozhen Hu, Cheng Tan, Jiangbin Zheng, Siyuan Li, and Stan Z. Li. Understanding the limitations of deep models for molecular property prediction: Insights and solutions. In A. Oh, T. Naumann, A. Globerson, K. Saenko, M. ...

2023
[18]

MaskMol: Knowledge-guided molecular image pre-training framework for activity cliffs with pixel masking

Zhixiang Cheng, Hongxin Xiang, Pengsen Ma, Li Zeng, Xin Jin, Xixi Yang, Jianxin Lin, Yang Deng, Bosheng Song, Xinxin Feng, Changhui Deng, and Xiangxiang Zeng. MaskMol: Knowledge-guided molecular image pre-training framework for activity cliffs with pixel masking. bioRxiv, 2024...

2024 doi
[19]

Prediction of activity cliffs on the basis of images using convolutional neural networks

Javed Iqbal, Martin V ogt, and Jürgen Bajorath. Prediction of activity cliffs on the basis of images using convolutional neural networks. Journal of Computer-Aided Molecular Design , pages 1–8, 2021. URL https://doi.org/10.1007/s10822-021-00380-y

2021 doi
[21]

Prediction of activity cliffs using condensed graphs of reaction representations, descriptor recombination, support vector machine classification, and support vector regression

Dragos Horvath, Gilles Marcou, Alexandre Varnek, Shilva Kayastha, Antonio de la Vega de León, and Jürgen Bajorath. Prediction of activity cliffs using condensed graphs of reaction representations, descriptor recombination, support vector machine classification, and support vec...

2016 doi
[22]

Exploring qsar models for activity- cliff prediction

Michael Dablander, Thomas Hanser, Renaud Lambiotte, et al. Exploring qsar models for activity- cliff prediction. Journal of Cheminformatics , 15:47, 2023. doi: 10.1186/s13321-023-00708-w. URL https://doi.org/10.1186/s13321-023-00708-w

2023 doi
[23]

Exposing the limitations of molecular machine learning with activity cliffs

Derek van Tilborg, Alisa Alenicheva, and Francesca Grisoni. Exposing the limitations of molecular machine learning with activity cliffs. Journal of Chemical Information and Modeling , 62(23):5938–5951,
[24]

Metric Learning: A Survey

Brian Kulis. Metric Learning: A Survey. F oundations and Trends in Machine Learning, 5(4):287–364,
[25]

Learning to Rank for Information Retrieval

Tie-Yan Liu. Learning to Rank for Information Retrieval . Springer-Verlag Berlin Heidelberg, Berlin, Heidelberg, 1st edition, 2011. URL https://doi.org/10.1007/978-3-642-14267-3 . 6

2011 doi
[26]

URL https://doi.org/10.1021/acs.jcim.2c01073

doi
[27]

Few-shot learning for low-data drug discovery.Journal of Chemical Information and Modeling, 63(1):27–42, 2023

Daniel Vella and Jean-Paul Ebejer. Few-shot learning for low-data drug discovery.Journal of Chemical Information and Modeling, 63(1):27–42, 2023. URL https://doi.org/10.1021/acs.jcim.2c00779

2023 doi
[28]

FS-Mol: A few-shot learning dataset of molecules

Megan Stanley, John Bronskill, Krzysztof Maziarz, Hubert Misztela, Jessica Lanini, Marwin Segler, Nadine Schneider, and Marc Brockschmidt. FS-Mol: A few-shot learning dataset of molecules. In J. Vanschoren and S. Yeung, editors, Proceedings of the Neural Information Processing...

2021
[29]

Implicitly guided design with PropEn: Match your data to follow the gradient, 2024

Nataša Tagasovska, Vladimir Gligorijevi ´c, Kyunghyun Cho, and Andreas Loukas. Implicitly guided design with PropEn: Match your data to follow the gradient, 2024. URL https://arxiv.org/abs/ 2405.18075

2024 arXiv
[30]

Pappu, and Vijay Pande

Han Altae-Tran, Bharath Ramsundar, Aneesh S. Pappu, and Vijay Pande. Low data drug discovery with one-shot learning. ACS Central Science, 3(4):283–293, 2017. URL https://doi.org/10.1021/ acscentsci.6b00367

2017
[31]

Twin neural network regression is a semi- supervised regression algorithm

Sebastian J Wetzel, Roger G Melko, and Isaac Tamblyn. Twin neural network regression is a semi- supervised regression algorithm. Machine Learning: Science and Technology, 3(4):045007, 2022. URL http://dx.doi.org/10.1088/2632-2153/ac9885

2022 doi
[32]

Pairwise difference learning for classification, 2024

Mohamed Karim Belaid, Maximilian Rabus, and Eyke Hüllermeier. Pairwise difference learning for classification, 2024. URL https://arxiv.org/abs/2406.20031

2024 arXiv
[33]

Burrill, Enrique R

Michael Tynes, Wenhao Gao, Daniel J. Burrill, Enrique R. Batista, Danny Perez, Ping Yang, and Nicholas Lubbers. Pairwise difference regression: A machine learning meta-algorithm for improved prediction and uncertainty quantification in chemical search. Journal of Chemical Info...

2021 doi
[34]

Twin neural network regression

Sebastian Johann Wetzel, Kevin Ryczko, Roger Gordon Melko, and Isaac Tamblyn. Twin neural network regression. Applied AI Letters, 3(4), 2022. URL http://dx.doi.org/10.1002/ail2.78

2022 doi
[35]

Finding the most potent compounds using active learning on molecular pairs

Zachary Fralish and Daniel Reker. Finding the most potent compounds using active learning on molecular pairs. Beilstein Journal of Organic Chemistry , 20:2152–2162, 2024. URL https://doi.org/10.3762/ bjoc.20.185

2024
[36]

Random forests

Leo Breiman. Random forests. Machine Learning, 45(1):5–32, 2001. URL https://doi.org/10. 1023/A:1010933404324

2001
[37]

XGBoost: A scalable tree boosting system

Tianqi Chen and Carlos Guestrin. XGBoost: A scalable tree boosting system. In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining , KDD ’16, pages 785–794, New York, NY , USA, 2016. ACM. URLhttp://doi.acm.org/10.1145/2939672.2939785

2016
[38]

DeepDelta: predicting ADMET improve- ments of molecular derivatives with deep learning

Zachary Fralish, Ashley Chen, Paul Skaluba, and Daniel Reker. DeepDelta: predicting ADMET improve- ments of molecular derivatives with deep learning. Journal of Cheminformatics, 15(1):101, 2023. URL https://doi.org/10.1186/s13321-023-00769-x

2023 doi
[39]

RDKit: Open-source cheminformatics software, 2016

Greg Landrum. RDKit: Open-source cheminformatics software, 2016. URL https://github.com/ rdkit/rdkit/releases/tag/Release_2016_09_4

2016
[40]

Strategies for pre-training graph neural networks

Weihua Hu*, Bowen Liu*, Joseph Gomes, Marinka Zitnik, Percy Liang, Vijay Pande, and Jure Leskovec. Strategies for pre-training graph neural networks. InInternational Conference on Learning Representations,
[41]

Prin- cipal neighbourhood aggregation for graph nets

Gabriele Corso, Luca Cavalleri, Dominique Beaini, Pietro Liò, and Petar Velickovic. Prin- cipal neighbourhood aggregation for graph nets. In Proceedings of the 34th International Conference on Neural Information Processing Systems , NIPS ’20, Red Hook, NY , USA,
[42]

Extended-connectivity fingerprints

David Rogers and Mathew Hahn. Extended-connectivity fingerprints. Journal of Chemical Information and Modeling, 50(5):742–754, 2010. URL https://doi.org/10.1021/ci100050t

2010 doi
[43]

Williams, Carl Underkoffler, Ryan Pederson, Narbe Mardirossian, Ian Watson, and John Parkhill

Benjamin Kaufman, Edward C. Williams, Carl Underkoffler, Ryan Pederson, Narbe Mardirossian, Ian Watson, and John Parkhill. COATI: Multimodal contrastive pretraining for representing and traversing chemical space. Journal of Chemical Information and Modeling , 64(4):1145–1157, ...

2024 doi
[44]

C Lim, and Prudencio Tossou

Emmanuel Noutahi, Cristian Gabellini, Michael Craig, Jonathan S. C Lim, and Prudencio Tossou. Gotta be SAFE: A new framework for molecular design, 2023. URL https://arxiv.org/abs/2310.10773

2023 arXiv
[45]

Uni-Mol: A universal 3D molecular representation learning framework

Gengmo Zhou, Zhifeng Gao, Qiankun Ding, Hang Zheng, Hongteng Xu, Zhewei Wei, Linfeng Zhang, and Guolin Ke. Uni-Mol: A universal 3D molecular representation learning framework. In The Eleventh International Conference on Learning Representations , 2023. URL https://openreview.n...

2023
[46]

Systematic benchmark of substructure search in molecular graphs - from Ullmann to VF2

Hans-Christian Ehrlich and Matthias Rarey. Systematic benchmark of substructure search in molecular graphs - from Ullmann to VF2. Journal of Cheminformatics, 4(1):13, 2012. URL https://doi.org/10. 1186/1758-2946-4-13

2012
[47]

URL https://proceedings.neurips.cc/paper/2020/file/ 99cad265a1768cc2dd013f0e740300ae-Paper.pdf

Curran Associates Inc. URL https://proceedings.neurips.cc/paper/2020/file/ 99cad265a1768cc2dd013f0e740300ae-Paper.pdf

2020
[48]

Molecular contrastive learning of representations via graph neural networks

Yuyang Wang, Jianren Wang, Zhonglin Cao, and Amir Barati Farimani. Molecular contrastive learning of representations via graph neural networks. Nature Machine Intelligence, 4(3):279–287, 2022. URL http://dx.doi.org/10.1038/s42256-022-00447-x . 7

2022 doi
[53]

Pedregosa, G

F. Pedregosa, G. Varoquaux, A. Gramfort, V . Michel, B. Thirion, O. Grisel, M. Blondel, P. Prettenhofer, R. Weiss, V . Dubourg, J. Vanderplas, A. Passos, D. Cournapeau, M. Brucher, M. Perrot, and E. Duchesnay. Scikit-learn: Machine learning in Python. Journal of Machine Learni...
[2011]

A Appendix A.1 Models We used the following models to evaluate the effectiveness of the SQRL approach: Baselines

URL http://jmlr.org/papers/v12/pedregosa11a.html. A Appendix A.1 Models We used the following models to evaluate the effectiveness of the SQRL approach: Baselines. All baseline models were trained on top of Morgan count fingerprints of size 2048, radius 2, and including chiral...

2000
[2013]

URL http://dx.doi.org/10.1561/2200000019

doi
[2019]

URL https://arxiv.org/abs/1907.11692

1907 arXiv
[2020]

URL https://openreview.net/forum?id=HJlWWJSFDH
[2022]

URL https://arxiv.org/abs/2102.09844

arXiv

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.

[1] [1]

Geometric deep learning on molecular repre- sentations

Kenneth Atz, Francesca Grisoni, and Gisbert Schneider. Geometric deep learning on molecular repre- sentations. Nature Machine Intelligence, 3(12):1023–1032, 2021. URL https://doi.org/10.1038/ s42256-021-00418-8

work page 2021

[2] [2]

Schoenholz, Patrick F

Justin Gilmer, Samuel S. Schoenholz, Patrick F. Riley, Oriol Vinyals, and George E. Dahl. Neural message passing for quantum chemistry. In Proceedings of the 34th International Conference on Machine Learning - V olume 70, ICML’17, pages 1263–1272, 2017. URL https://proceedings.mlr.press/ v70/gilmer17a/gilmer17a.pdf

work page 2017

[3] [3]

Analyzing learned molecular representations for property prediction

Kevin Yang, Kyle Swanson, Wengong Jin, Connor Coley, Philipp Eiden, Hua Gao, Angel Guzman-Perez, Timothy Hopper, Brian Kelley, Miriam Mathea, Andrew Palmer, V olker Settels, Tommi Jaakkola, Klavs Jensen, and Regina Barzilay. Analyzing learned molecular representations for property prediction. Journal of Chemical Information and Modeling , 59(8):3370–3388,...

work page 2019

[4] [4]

Pushing the boundaries of molecular representation for drug discovery with the graph attention mechanism

Zhaoping Xiong, Dingyan Wang, Xiaohong Liu, Feisheng Zhong, Xiaozhe Wan, Xutong Li, Zhaojun Li, Xiaomin Luo, Kaixian Chen, Hualiang Jiang, and Mingyue Zheng. Pushing the boundaries of molecular representation for drug discovery with the graph attention mechanism. Journal of Medicinal Chemistry , 63 (16):8749–8760, 2020. URL https://doi.org/10.1021/acs.jme...

doi:10.1021/acs.jmedchem.9b00959 2020

[5] [5]

Spherical message passing for 3D molecular graphs

Yi Liu, Limei Wang, Meng Liu, Yuchao Lin, Xuan Zhang, Bora Oztekin, and Shuiwang Ji. Spherical message passing for 3D molecular graphs. In International Conference on Learning Representations, 2022. URL https://openreview.net/forum?id=givsRXsOt9r

work page 2022

[6] [6]

K. T. Schütt, P.-J. Kindermans, H. E. Sauceda, S. Chmiela, A. Tkatchenko, and K.-R. Müller. SchNet: A continuous-filter convolutional neural network for modeling quantum interactions. In Proceedings of the 31st International Conference on Neural Information Processing Systems , NIPS’17, pages 992–1002, Red Hook, NY , USA, 2017. Curran Associates Inc. URLh...

work page 2017

[7] [7]

E(n) equivariant graph neural networks,

Victor Garcia Satorras, Emiel Hoogeboom, and Max Welling. E(n) equivariant graph neural networks,

[8] [8]

SMILES-BERT: Large scale unsupervised pre-training for molecular property prediction

Sheng Wang, Yuzhi Guo, Yuhong Wang, Hongmao Sun, and Junzhou Huang. SMILES-BERT: Large scale unsupervised pre-training for molecular property prediction. In Proceedings of the 10th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics , BCB ’19, pages 429–436, New York, NY , USA, 2019. Association for Computing Machi...

arXiv 2019