Pith. sign in

REVIEW 4 major objections 5 minor 35 references

Learning Cross-Domain Representations for Transferable Drug Perturbations on Single-Cell Transcriptional Responses

T0 review · 4 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read Cross-domain swapping lets a model predict single-cell responses to drugs and gene knockouts it has never seen.

desk verdict A sensible extension of cycleCDR with a cross-transfer loss, but the drug-response evaluation is transductive, not zero-shot, and the headline overclaims. read the letter →

arxiv 2412.19228 v2 pith:NYGQY3XR submitted 2024-12-26 cs.LG cs.AI

Hui Liu , Shikai Jin This is my paper · ORCID
classification cs.LGcs.AI
keywords single-celltranscriptomicsdrugresponsepredictioncross-domaintransferlearningdisentangledrepresentationlinearadditivelatentspaceperturbationcombinatorialgeneticperturbationsphenotypicdiscovery
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 single-cell transcriptional responses to drugs and genetic perturbations can be predicted for unseen perturbations by learning a perturbation representation that transfers across cellular contexts. XTransferCDR encodes each perturbed expression profile into a basal cell state and a perturbation vector, then forces the model to reconstruct the paired expression profile after the two perturbation vectors are swapped. This cross-transfer constraint is what makes the perturbation representations transferable. On three benchmark families, including sci-Plex drug screens, Perturb-seq single-gene knockouts, and combinatorial genetic perturbations, the reported R2, explained variance, and Pearson correlation beat the comparison methods, with the largest gains on differentially expressed genes. If the claim holds, drug screening could query many more drugs and drug combinations in silico before committing to wet-lab experiments.

What carries the argument

The load-bearing mechanism is the cross-transfer constraint encoded in the loss $L_{cross}$: the decoder must turn the sum of perturbation $P^{(a)}$ and basal state $S^{(b)}$ into expression profile $X^{(a)}$, and vice versa. The loss is the mean squared error between the cross-transferred reconstructions and the original perturbed profiles. This is supported by an orthogonality loss $L_{orth}$ that separates perturbation from basal information, a similarity loss $L_{sim}$ that aligns basal states extracted from two perturbations, and two reconstruction losses; the total objective is $L = L_{sim} + L_{orth} + L^{(1)}_{reco} + L^{(2)}_{reco} + L_{cross}$. The linear-additivity rule, that the effect of a perturbation on the latent cellular state is a vector addition to the basal state, is what justifies both the cross-transfer reconstruction and the combinatorial prediction $S + P^{(a)} + P^{(b)}$.

What would settle it

Take a set of gene pairs with documented epistatic or non-additive interactions, train XTransferCDR on single-gene knockouts only, and compare its predicted double-knockout profiles $S + P^{(a)} + P^{(b)}$ to actual measured profiles; if DEG-level R2 collapses specifically on the epistatic pairs while staying high on additive ones, the linear-additivity assumption is falsified.

Watch

Extended reading notes

Core claim

XTransferCDR's central discovery is that a cross-transfer reconstruction loss is sufficient to learn perturbation representations that remain valid when moved to a different basal state. Given expression profiles $X^{(a)}$ and $X^{(b)}$ from two perturbations applied to the same cell context, two encoders extract basal states $S^{(a)}$, $S^{(b)}$ and perturbation vectors $P^{(a)}$, $P^{(b)}$. The model reconstructs the original profiles from $S+P$ and, crucially, also reconstructs $X^{(a)}$ from $S^{(b)}+P^{(a)}$ and $X^{(b)}$ from $S^{(a)}+P^{(b)}$, with a shared decoder. Under the stated assumption that perturbations add linearly to the latent cellular state, this forces $P$ to encode only the perturbation, not the context; after training, a novel drug's response is predicted by adding its extracted $P$ to the basal state of a target cell population and decoding the sum. The paper evaluates this on held-out drugs and gene knockouts, and also predicts dual-gene knockout profiles by summing two single-gene perturbation vectors, with R2 on differentially expressed genes rising from 0.166 for the baseline to 0.837.

Load-bearing premise

The model's transferability rests on the assumption that a perturbation's effect is a vector that adds linearly to the cell's latent basal state, so that swapping or summing vectors in the learned space corresponds to actually applying the perturbations.

Editorial extensions

If this is right

  • A model trained on existing single-cell drug screens can predict expression profiles for held-out drugs, since the test protocol separated drugs by identity between training and test.
  • The same learned perturbation vectors predict single-gene knockout responses in K562 and RPE-1 cells, with DEG-level R2 of 0.68 and 0.78 respectively, roughly double the best comparison method.
  • Summing two single-gene perturbation vectors predicts dual-gene knockout profiles: on the ten held-out combinatorial perturbations, DEG-level R2 reaches 0.837 versus 0.166 for the baseline.
  • Ablations show the cross-transfer loss is the main driver: removing it drops all-gene R2 on sci-Plex3 from 0.817 to 0.660 and DEG-level R2 from 0.626 to 0.183.
  • Because perturbation vectors live in a shared latent space, the model offers an interpretable arithmetic: similar drugs should have similar $P$ vectors, which can be compared directly.

Reading between the lines

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

  • A testable extension the authors do not run: systematically scan gene pairs with documented epistatic or non-additive effects and compare $S + P^{(a)} + P^{(b)}$ predictions to measured double-perturbation profiles, which would show where the linear-additivity representation breaks.
  • Because the model predicts only from expression profiles, it should also propose perturbation vectors for entirely unseen cell types by swapping in that cell type's basal state, a direct extension to patient-derived or tissue-specific contexts.
  • The same framework could be pointed at drug combination synergy: treating two drugs' vectors as additive would allow ranking candidate combinations by the distance between predicted and desired phenotype before wet-lab screening.
  • Interpreting $P$ vectors as drug directions lets researchers cluster drugs by mechanism of action without pathway knowledge, and could flag off-target effects when a drug's vector points along an unexpected biological axis.
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.

Referee Report

4 major / 5 minor

Summary. The paper proposes XTransferCDR, a generative framework that disentangles perturbed single-cell expression profiles into a basal-state representation and a perturbation representation via two encoders, then enforces a linear additivity constraint in the latent space and a cross-transfer reconstruction loss to learn transferable perturbation representations. The authors evaluate on sci-Plex drug-response data, single-gene Perturb-seq data, and combinatorial perturbation data, reporting improved R2, EV, and PCC over several baselines, and conclude that their method advances phenotypic drug discovery by predicting responses to novel drugs and perturbations.

Significance. If the claimed ability to predict transcriptional responses to unseen drugs were real, the method would be a useful tool for drug screening and for interpreting perturbation effects. The combinatorial perturbation experiment (Table 4) is a genuine extrapolation, and the cross-domain disentanglement idea is a reasonable contribution to the representation-learning toolbox. The paper also ships a public code repository, which aids reproducibility. However, the central evaluation protocol for the drug and single-gene experiments feeds the target perturbed expression profile into the model at test time, making those experiments circular and invalidating the headline claims of transferable prediction. The net significance is therefore substantially below what the abstract and introduction assert.

major comments (4)
  1. [Experimental Setup] The test-time protocol is circular for the drug and single-gene experiments. The paper states: “In the testing stage, the perturbed expression profile by a perturbation ‘unseen’ in training stage was fed into the encoder to extract its perturbation representation, which is then combined with the basal state derived from corresponding unperturbed expression profiles.” This means the model receives the actual perturbed profile (the prediction target) as input to Ep, and the decoder reconstructs a function of that target. The comparisons in Tables 1–3 against chemCPA, GEARS, and cycleCDR are therefore not apples-to-apples: those baselines do not receive the test drug’s perturbed profile and are being asked to predict it, while XTransferCDR is doing transductive reconstruction. The high R2/PCC/EV values are largely a measure of the autoencoder’s ability to compress and reproduce its input, not of transferable prediction. This undermines the abstract’s central claim that the model “achieved better performance than current state-of-the-art methods” for transferable drug perturbations.
  2. [XTransferCDR / Representation Disentanglement] The architecture cannot, as described, produce a perturbation representation for a truly unseen drug without already having that drug’s perturbed expression profile. The only path to P(a) is Ep(X(a)), i.e., the encoder requires the perturbed expression profile itself. For a novel drug, by definition, no such profile is available. Thus the method is not a predictor of cellular responses to novel drugs; it is a profile autoencoder/reconstructor. The claim in “Experimental Setup” that the strategy evaluates “the model’s ability to generalize to novel drugs” is contradicted by the protocol. To claim zero-shot drug prediction, the paper would need a separate drug representation source (e.g., a molecular-structure encoder like ChemCPA’s) or a method to derive P for an unseen drug from training drugs; neither is present.
  3. [Evaluation on Combinatorial Genetic Perturbations] Table 4 is presented as a robust comparison with GEARS (“these dual-gene perturbations were also chosen for testing in the GEARS study”), but the table only reports Baseline and XTransferCDR metrics. No GEARS results are shown or discussed beyond that sentence. Moreover, this experiment is the only non-circular evaluation, yet it relies entirely on the unvalidated linear-additivity assumption: the model predicts S + P(a) + P(b) for a dual-gene perturbation from single-gene perturbation vectors. The cross-transfer loss in Eq. (5) is built on the same additivity assumption, so the strong Table 4 R2 values are partially an in-sample fit to that assumption. The paper does not provide direct evidence that dual-gene transcriptional effects are additive in latent space (e.g., by comparing predicted dual effects against a non-additive baseline or by quantifying the error of additivity in held-out combinations).
  4. [Model Ablations] The ablation study in Table 5 uses the same circular test protocol as Tables 1–3. Since the model receives the target profile at test time, the observed improvements from adding Lcross and the other losses could reflect better reconstruction of the given target rather than better transfer of perturbation representations. The ablations therefore do not provide independent support for the claim that the cross-transfer constraint promotes generalizable representations.
minor comments (5)
  1. [Representation Disentanglement, Eq. (1)] The notation in Eq. (1) is unclear: P_i^(a) and S_i^(a) are presumably vectors, so their dot product is a scalar, and the squared Frobenius norm reduces to an absolute square. The authors should rewrite this as ‖(P_i^(a))^T S_i^(a)‖^2 or clearly define the operation.
  2. [Representation Disentanglement, Eq. (2)] The text says the basal states “should be aligned to each other as much as possible” but then writes “we try to minimize the similarity between them.” Since the objective is KL divergence, minimizing it makes the distributions more similar. The sentence should say “minimize the divergence” or “maximize the similarity.”
  3. [Experimental Setup] The description of the Basline model is ambiguous: “calculates performance metrics directly using the unperturbed and actual perturbed expression profiles.” If the baseline is simply predicting the unperturbed profile for every perturbation, this should be stated explicitly. If it is using the actual perturbed profile as its own prediction, the R2 values in Tables 1–3 would be nonsensical.
  4. [Experimental Setup] The paper does not report error bars or significance tests for any of the tables. Since the datasets are split once at random, all reported numbers are point estimates from a single split; without multiple seeds or confidence intervals, it is hard to assess whether the gaps between methods are meaningful.
  5. [Introduction] The phrase “to our knowledge, this is the first time that cross-domain disentanglement representation learning has been proposed to model perturbation-induced cellular responses” is overclaimed. CycleCDR, which is cited, already uses cycle-consistency for cross-domain perturbation transfer; the authors should temper this claim and position their contribution more precisely relative to cycleCDR.

Circularity Check

1 steps flagged · score 6.0 of 10

Held-out drug's actual perturbed profile is fed to the encoder at test time, so Tables 1-3 measure target reconstruction, not transferable prediction.

  1. self definitional [Experiments, Experimental Setup (testing-stage protocol); see also Eq. (4) and Eq. (5)]
    "In the testing stage, the perturbed expression profile by a perturbation “unseen” in training stage was fed into the encoder to extract its perturbation representation, which is then combined with the basal state derived from corresponding unperturbed expression profiles. This testing strategy was intentionally designed to evaluate the model’s ability to generalize to novel drugs when predicting cellular transcriptional responses."

    For a held-out drug, the test input to Ep is the drug's actual perturbed expression profile, which is exactly the quantity the paper reports as 'predicted'. The model computes P = Ep(X_perturbed) and outputs D(S + P); the training objective already optimizes this same reconstruction path, e.g. Lcross in Eq. (5) trains D(S(b)+P(a)) to reproduce X(a). Thus the reported R2, EV, and PCC in Tables 1-3 measure how well an autoencoder reconstructs its own input through a bottleneck, not how well the model predicts an unseen drug response from the drug's identity or structure. The 'unseen' label refers only to the drug's absence from training; the target profile itself is supplied to the model, so the claimed generalization to novel drugs is not evaluated.

full rationale

The one concrete circularity is the held-out drug test protocol: the actual perturbed profile is fed to Ep, so the 'prediction' is a reconstruction of the target. This undermines the Tables 1-3 superiority claims and the abstract's claim of transferable prediction for unseen drugs and genetic perturbations, because the model is given target information that the comparison methods (chemCPA, GEARS-style baselines) do not receive. The combinatorial experiment (Table 4) is a genuine extrapolation: dual-gene profiles are predicted from single-gene representations via S + P_a + P_b, with no dual-gene target fed to the encoders; its validity rests on the stated linear-additivity assumption, which is an assumption rather than a circular derivation. Self-citations to cycleCDR are not load-bearing here: the linear-additivity postulate is visibly imported as a modeling assumption, not established by the citation. The training objective itself is self-contained and the disentanglement losses are not definitionally equal to the target. Hence the circularity is partial and concentrated in the evaluation protocol rather than in the mathematical derivation, giving a score of 6.

Assumptions & free parameters 4 free parameters · 4 assumptions · 2 invented entities

The method rests on several domain assumptions (linear additivity, basal state invariance) that are not independently validated. The main free parameters are the loss weights and architecture choices, which are not fully specified. The latent representations S and P are invented constructs without external evidence, and their usefulness is demonstrated only within the model's own reconstruction framework.

free parameters (4)
  • Loss weighting coefficients for Lsim, Lorth, Lreco(1), Lreco(2), Lcross = unspecified
    The paper states the loss terms were 'appropriately weighted' but gives no exact values; these weights are tuned on validation and materially affect the results.
  • Bottleneck dimension = 128
    The latent dimension between encoder and decoder is set to 128, an architectural choice that affects the capacity for disentanglement.
  • Network architecture hyperparameters (layer sizes, dropout, learning rate, epochs) = four layers of sizes 1024, 512, 256, 128; dropout 0.2; learning rate 2e-4; 60 epochs
    These hyperparameters are chosen by hand or tuned on the validation set; they are not derived from first principles.
  • DEG threshold and top-50 DEG count for evaluation metrics = log2|fold change| >= 1; top 50 DEGs
    The definition of differentially expressed genes and the choice to report metrics on the top 50 DEGs are arbitrary thresholds that influence the reported performance numbers.
assumptions (4)
  • domain assumption Perturbation effects follow a linear additive rule in the latent space.
    Stated in the 'Expression Profile Reconstruction' subsection; the entire cross-transfer and combinatorial prediction scheme relies on this hypothesis.
  • domain assumption The basal state is invariant across different perturbations applied to the same cellular context.
    Required for the similarity loss Lsim (Eq. 2) and for the validity of swapping basal states in the cross-transfer loss.
  • domain assumption Orthogonality between the basal state and perturbation representation is achievable and beneficial.
    Enforced via Lorth (Eq. 1); the model assumes this disentanglement is learnable and improves transferability.
  • domain assumption The decoder can map the additive combination of basal and perturbation representations back to expression profiles.
    Implicitly assumed when defining Lreco(2) and Lcross; the decoder must treat the sum as a valid latent code.
invented entities (2)
  • Basal state representation S
    purpose: A latent vector intended to capture the unperturbed cellular state of a cell, used to reconstruct the unperturbed profile and to combine with perturbation vectors.
    S is a learned latent variable with no external validation; the paper does not demonstrate that S corresponds to any measurable biological quantity.
  • Perturbation representation P
    purpose: A latent vector intended to encode the effect of a drug or genetic perturbation, transferable across cellular contexts.
    P is a learned latent variable. The paper provides no independent evidence that P captures a real, transferable biological signal; its validity is only assessed indirectly through reconstruction performance.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Learning Cross-Domain Representations for Transferable Drug Perturbations on Single-Cell Transcriptional Responses." pith.science (2026). https://pith.science/paper/NYGQY3XR

@misc{pith2026241219228,
  author       = {Pith},
  title        = {Pith review of: Learning Cross-Domain Representations for Transferable Drug Perturbations on Single-Cell Transcriptional Responses},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NYGQY3XR}},
  note         = {Machine review of arXiv:2412.19228}
}
read the original abstract

Phenotypic drug discovery has attracted widespread attention because of its potential to identify bioactive molecules. Transcriptomic profiling provides a comprehensive reflection of phenotypic changes in cellular responses to external perturbations. In this paper, we propose XTransferCDR, a novel generative framework designed for feature decoupling and transferable representation learning across domains. Given a pair of perturbed expression profiles, our approach decouples the perturbation representations from basal states through domain separation encoders and then cross-transfers them in the latent space. The transferred representations are then used to reconstruct the corresponding perturbed expression profiles via a shared decoder. This cross-transfer constraint effectively promotes the learning of transferable drug perturbation representations. We conducted extensive evaluations of our model on multiple datasets, including single-cell transcriptional responses to drugs and single- and combinatorial genetic perturbations. The experimental results show that XTransferCDR achieved better performance than current state-of-the-art methods, showcasing its potential to advance phenotypic drug discovery.

Figures

Figures reproduced from arXiv: 2412.19228 by the authors.

Figure 1
Figure 1. Illustrative diagram of the proposed XTransferCDR framework. Two encoders [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Performance evaluation on sci-plex single-cell [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Performance evaluation on single-cell transcrip [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Predicted single-cell expression levels of top 20 [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

35 extracted references · 31 canonical work pages

  1. [1]

    , " * write output.state after.block = add.period write newline

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.a...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  3. [3]

    F.; Islam, A

    Aissa, A. F.; Islam, A. B.; Ariss, M. M.; Go, C. C.; Rader, A. E.; Conrardy, R. D.; Gajda, A. M.; Rubio-Perez, C.; Valyi-Nagy, K.; Pasquinelli, M.; et al. 2021. Single-cell transcriptional changes associated with drug tolerance and response to combination therapies in cancer. Nature communications, 12(1): 1628

    work page 2021

  4. [4]

    Bousmalis, K.; Trigeorgis, G.; Silberman, N.; Krishnan, D.; and Erhan, D. 2016. Domain separation networks. Advances in neural information processing systems, 29

    2016

  5. [5]

    Chawla, S.; Rockstroh, A.; Lehman, M.; Ratther, E.; Jain, A.; Anand, A.; Gupta, A.; Bhattacharya, N.; Poonia, S.; Rai, P.; et al. 2022. Gene expression based inference of cancer drug sensitivity. Nature communications, 13(1): 5680

    work page 2022

  6. [6]

    Ding, S.; Chen, X.; and Shen, K. 2020. Single‐cell RNA sequencing in breast cancer: Understanding tumor heterogeneity and paving roads to individualized therapy. Cancer Communications, 329–344

    work page 2020

  7. [7]

    Gao, L.-G.; Yang, M.-Y.; and Wang, J.-X. 2021. Collaborative matrix factorization with soft regularization for drug-target interaction prediction. Journal of Computer Science and Technology, 36: 310--322

    work page 2021

  8. [8]

    Gonzalez-Garcia, A.; Weijer, J.; and Bengio, Y. 2018. Image-to-image translation for cross-domain disentanglement. Neural Information Processing Systems,Neural Information Processing Systems

    work page 2018

Show all 35 references
  1. [9]

    A.; and Lönnberg, T

    Haque, A.; Engel, J.; Teichmann, S. A.; and Lönnberg, T. 2017. A practical guide to single-cell RNA-sequencing for biomedical research and clinical applications. Genome Medicine

    2017

  2. [10]

    He, D.; Liu, Q.; Wu, Y.; and Xie, L. 2022. A context-aware deconfounding autoencoder for robust prediction of personalized clinical drug response from cell-line compound screening. Nature Machine Intelligence, 4(10): 879--892

    2022

  3. [11]

    A.; Scotti, L.; and Scotti, M

    Herrera-Acevedo, C.; Perdomo-Madrigal, C.; de Sousa Luis, J. A.; Scotti, L.; and Scotti, M. T. 2022. Drug Discovery Paradigms: Target-Based Drug Discovery, 1--24. Cham: Springer International Publishing. ISBN 978-3-030-95895-4

    2022

  4. [12]

    Hetzel, L.; Boehm, S.; Kilbertus, N.; G \"u nnemann, S.; Theis, F.; et al. 2022. Predicting cellular responses to novel drug perturbations at a single-cell resolution. Advances in Neural Information Processing Systems, 35: 26711--26722

    2022

  5. [13]

    Huang, W.; and Liu, H. 2024. Predicting single-cell cellular responses to perturbations using cycle consistency learning. Bioinformatics, 40(Supplement\_1): i462--i470

    2024

  6. [14]

    Jia, P.; Hu, R.; Pei, G.; Dai, Y.; Wang, Y.-Y.; and Zhao, Z. 2021. Deep generative neural network for accurate drug response imputation. Nature Communications, 12(1): 1740

    2021

  7. [15]

    H.; Mitchell, C

    Jones, L. H.; Mitchell, C. A.; Loberg, L.; Pavkovic, M.; Rao, M.; Roberts, R.; Stamp, K.; Volak, L.; Wittwer, M. B.; and Pettit, S. 2022. Targeted protein degraders: a call for collective action to advance safety assessment. Nature Reviews Drug Discovery, 21(6): 401–402

    2022

  8. [16]

    Lee, S.; Cho, S.; and Im, S. 2021. DRANet: Disentangling Representation and Adaptation Networks for Unsupervised Cross-Domain Adaptation. In 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

    2021

  9. [17]

    M.; Yuan, J.; and Sims, P

    Levitin, H. M.; Yuan, J.; and Sims, P. A. 2018. Single-Cell Transcriptomic Analysis of Tumor Heterogeneity. Trends in Cancer, 4(4): 264–268

    2018

  10. [18]

    Lim, W. A. 2022. The emerging era of cell engineering: Harnessing the modularity of cells to program complex biological function. Science, 848–852

    2022

  11. [19]

    K.; De Donno, C.; Ji, Y.; Ibarra, I

    Lotfollahi, M.; Susmelj, A. K.; De Donno, C.; Ji, Y.; Ibarra, I. L.; Wolf, F. A.; Yakubova, N.; Theis, F. J.; and Lopez-Paz, D. 2021. Predicting cellular responses to complex perturbations in high-throughput screens. BioRxiv, 2021--04

    2021

  12. [20]

    S.; Khade, P

    Madhukar, N. S.; Khade, P. K.; Huang, L.; Gayvert, K.; Galletti, G.; Stogniew, M.; Allen, J. E.; Giannakakou, P.; and Elemento, O. 2019. A Bayesian machine learning approach for drug target identification using diverse data types. Nature communications, 10(1): 5221

    2019

  13. [21]

    Nguyen, A.-P.; Vasilaki, S.; and Mart \' nez, M. R. 2023. FLAN: feature-wise latent additive neural models for biological applications. Briefings in Bioinformatics, 24(3): bbad056

    2023

  14. [22]

    M.; Horlbeck, M

    Norman, T. M.; Horlbeck, M. A.; Replogle, J. M.; Ge, A. Y.; Xu, A.; Jost, M.; Gilbert, L. A.; and Weissman, J. S. 2019. Exploring genetic interaction manifolds constructed from rich single-cell phenotypes. Science, 786–793

    2019

  15. [23]

    Przybyla, L.; and Gilbert, L. A. 2022. A new era in functional genomics screens. Nature Reviews Genetics, 23(2): 89–103

    2022

  16. [24]

    M.; Saunders, R

    Replogle, J. M.; Saunders, R. A.; and et al. 2021. Mapping information-rich genotype-phenotype landscapes with genome-scale Perturb-seq. Cell, 185(14): 2559 -- 2575.e28

    2021

  17. [25]

    P.; Le, X.; Vijayan, R.; Hicks, J

    Robichaux, J. P.; Le, X.; Vijayan, R.; Hicks, J. K.; Heeke, S.; Elamin, Y. Y.; Lin, H. Y.; Udagawa, H.; Skoulidis, F.; Tran, H.; et al. 2021. Structure-based classification predicts drug response in EGFR-mutant NSCLC. Nature, 597(7878): 732--737

    2021

  18. [26]

    Roohani, Y.; Huang, K.; and Leskovec, J. 2022. GEARS: Predicting transcriptional outcomes of novel multi-gene perturbations. BioRxiv, 2022--07

    2022

  19. [27]

    A.; Zolotareva, O.; Collins, C

    Sharifi-Noghabi, H.; Harjandi, P. A.; Zolotareva, O.; Collins, C. C.; and Ester, M. 2021. Out-of-distribution generalization from labelled and unlabelled gene expression data for drug response prediction. Nature Machine Intelligence, 3(11): 962--972

    2021

  20. [28]

    R.; McFaline-Figueroa, J

    Srivatsan, S. R.; McFaline-Figueroa, J. L.; Ramani, V.; Saunders, L.; Cao, J.; Packer, J.; Pliner, H. A.; Jackson, D. L.; Daza, R. M.; Christiansen, L.; Zhang, F.; Steemers, F.; Shendure, J.; and Trapnell, C. 2020. Massively multiplex chemical transcriptomics at single-cell re...

    2020

  21. [29]

    M.; and et al

    Subramanian, A.; Narayan, R.; Corsello, S. M.; and et al. 2017. A next generation connectivity map: L1000 platform and the first 1,000,000 profiles. Cell, 171(6): 1437--1452.e17

    2017

  22. [30]

    Tran, V.-H.; and Huang, C.-C. 2019. Domain adaptation meets disentangled representation learning and style transfer. In 2019 IEEE International Conference on Systems, Man and Cybernetics (SMC), 2998--3005. IEEE

    2019

  23. [31]

    J.; Lightfoot, H.; Forbes, S.; Bindal, N.; Beare, D.; Smith, J

    Yang, W.; Soares, J.; Greninger, P.; Edelman, E. J.; Lightfoot, H.; Forbes, S.; Bindal, N.; Beare, D.; Smith, J. A.; Thompson, I. R.; et al. 2012. Genomics of Drug Sensitivity in Cancer (GDSC): a resource for therapeutic biomarker discovery in cancer cells. Nucleic acids resea...

    2012

  24. [32]

    Yofe, I.; Dahan, R.; and Amit, I. 2020. Single-cell genomic approaches for developing the next generation of immunotherapies. Nature Medicine, 26(2): 171–177

    2020

  25. [33]

    Yu, H.; and Welch, J. D. 2022. PerturbNet predicts single-cell responses to unseen chemical and genetic perturbations. bioRxiv, 2022--07

    2022

  26. [34]

    Zhang, C.; Lu, Y.; and Zang, T. 2022. CNN-DDI: a learning-based method for predicting drug--drug interactions using convolution neural networks. BMC bioinformatics, 23(1): 1--12

    2022

  27. [35]

    Zhu, J.-Y.; Park, T.; Isola, P.; and Efros, A. A. 2017. Unpaired image-to-image translation using cycle-consistent adversarial networks. In Proceedings of the IEEE international conference on computer vision, 2223--2232

    2017

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.