Pith. sign in

REVIEW 3 major objections 4 minor 44 references

Generating Synthetic Genotypes using Diffusion Models

T0 review · 3 major / 4 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 A conditional diffusion model can generate complete synthetic human genotypes whose use as training data recovers about 94% of real-data accuracy for ALS and 93% for population classification.

desk verdict A real scale advance in genotype embedding generation, but the paper overclaims 'complete synthetic genotypes' without ever decoding an embedding back to DNA. read the letter →

arxiv 2412.03278 v3 pith:EALESCFR submitted 2024-12-04 cs.CE

Philip Kenneweg , Raghuram Dandinasivara , Xiao Luo , Barbara Hammer , Alexander Schönhuth This is my paper · ORCID
classification cs.CE
keywords syntheticgenotypesdiffusionmodelsprivacy-preservinggenomicsgenome-wideassociationstudiesdeepgenerativeALSclassification1000GenomesPCAembedding
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 to be the first to generate complete, full-length synthetic human genotypes with a diffusion model, then expand them into DNA-level genomes by standard reference-based protocols. Its central assertion is that these synthetic genotypes are realistic and novel, not memorized copies of the training genomes, and that classifiers trained on them recover about 94% (ALS) and 93% (1000 Genomes populations) of the accuracy achieved on real data. The paper also claims that adding synthetic genotypes to small real training sets restores near-full accuracy, e.g., from about 71% to about 85% when only 5% of real ALS data is available. This matters because real whole-genome data are privacy-restricted and expensive to share; a synthetic replacement could let researchers distribute useful genomic training data without releasing sensitive genotypes.

What carries the argument

The mechanism is a diffusion model operating on per-gene PCA embeddings. Each gene's 5-100 SNP positions are compressed with principal component analysis into 1-8 components, zero-padded to length 8, and stacked across 18,279 genes (ALS) or 26,624 genes (1KG), producing real-valued inputs of shape $18432 \times 8$ or $26624 \times 8$. A U-Net-style denoiser learns to predict the Gaussian noise added in the forward diffusion process; a conditioning label $y$ (disease status or population) is injected to steer generation. The strongest variant combines a position-unbiased MLP U-Net with a 1D CNN U-Net through a learned weight $\lambda(t)$ that shifts emphasis from coarse structure to fine detail as noise decreases.

What would settle it

Decode a batch of synthetic embeddings by inverting the per-gene PCA, reconstructing ternary (or binary haplotype) genotype vectors, and compare the reconstructed genotypes to real ones in allele frequency, linkage disequilibrium, and classifier accuracy. If classifiers trained on the decoded synthetic genotypes lose most of the reported recovery rates, the claim of generating full-length synthetic genotypes is not established.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that a conditional diffusion model, trained in a compressed per-gene principal-component space, can sample novel full-length genotype profiles that carry the signal distinguishing ALS patients from controls and the signals separating 26 human populations. The best-performing generator, an MLP and 1D CNN U-Net hybrid with a learned time-dependent mixing weight, reaches recovery rates of 94.26% on the ALS task and 93.02% on the 1KG task when the same MLP classifier is trained on synthetic instead of real data. Nearest-neighbour adversarial accuracy stays near the 0.5 level that indicates the synthetic data cannot be distinguished from real data, and no synthetic sample matches a real sample in L1, L2, or cosine distance. From these results the authors conclude that synthetic genotypes are realistic, diverse, and suitable as training data for translational genetics, and that small real cohorts can be augmented with synthetic data to approach full-data performance.

Load-bearing premise

The load-bearing premise is that compressing each gene to a few principal components preserves every genetic signal needed for disease and population classification, because the model both generates and evaluates in this compressed space and never decodes synthetic embeddings back into actual SNP genotypes or DNA sequences.

Editorial extensions

If this is right

  • A researcher with only 5% of a real cohort can add synthetic genotypes and move ALS classifier accuracy from about 71% back to about 85%, and 1KG population accuracy from about 29% to about 84%.
  • Conditional generation means balanced disease-versus-control synthetic cohorts can be produced on demand, easing class-imbalance problems in small real datasets.
  • Access-restricted cohorts could be shared as synthetic data without releasing either the real genotypes or the trained generator, lowering the barrier for reproducing genomic studies.
  • The reported nearest-neighbour privacy-loss scores support novelty of the samples, within the limits of an empirical rather than a formal privacy guarantee.

Reading between the lines

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

  • Inferring beyond the paper: the absence of an inverse-PCA decoding step means the 'full-length genotype' claim is only as strong as the embedding's fidelity; a decoding-based evaluation would be the natural next test.
  • Inferring beyond the paper: because the generator is label-conditioned, a promising extension is generating synthetic genomes for under-represented populations or rare disease subgroups to rebalance training sets.
  • Inferring beyond the paper: the empirical privacy metric rules out exact copying but not inference attacks; a differentially private training procedure would be needed to turn privacy from an empirical property into a guarantee.
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

3 major / 4 minor

Summary. The paper proposes GeneticDiffusion, a class-conditioned diffusion model that generates synthetic human genotype profiles in a per-gene PCA embedding space. The authors train four generator architectures (MLP, CNN, MLP+CNN, Transformer) on ALS and 1000 Genomes data, then evaluate the synthetic data by training downstream classifiers on it and measuring recovery rates relative to classifiers trained on real data (Tables 2 and 6), by augmentation experiments with small amounts of real data (Table 3), and by nearest-neighbour adversarial accuracy / privacy-loss metrics (Table 4). The paper claims that this is the first diffusion model that generates complete synthetic human genotypes, which can be expanded to full-length DNA-level genomes using reference-based standard protocols.

Significance. If the central claim were fully established, the paper would be a significant contribution to privacy-preserving sharing of human genetic data: a generative model that produces realistic, diverse, full-length genotypes could enable downstream classifier training without releasing sensitive individual-level data. The paper has several strengths: the experimental protocol is mostly transparent, Tables 2 and 6 are internally consistent (e.g., 1KG MLP+CNN recovery 83.98/90.23 = 93.06%, matching the stated 93.02%), the architecture comparison is systematic, and the NNAA experiments address the important issue of training-data memorization. However, the central claim is currently supported only at the level of PCA embeddings, not at the level of actual genotypes or DNA, and the augmentation experiment contains a leakage confound. These issues are load-bearing for the paper's headline claims and require additional experiments rather than mere rewriting.

major comments (3)
  1. [Section 3.2, Fig. 1, Abstract] The paper never decodes a synthetic embedding back into a discrete SNP genotype or a DNA-level genome. All generation, evaluation, and classifier training occur in the per-gene PCA embedding space (R^18432x8 or R^26624x8), and the classifier test sets are also embeddings of real data. The abstract's claim of generating 'complete synthetic human genotypes' expandable to 'full-length, DNA-level genomes' is therefore not directly demonstrated. The <1% compression loss cited from Luo et al. (2023) concerns PCA reconstruction error on real data, not the validity of decoded synthetic samples; small continuous perturbations in embedding space can flip discrete allele calls, and the claim that non-SNP sites can be imputed by LD principles is asserted, not verified. To support the central claim, the authors should add an experiment that inverts synthetic embeddings to SNP genotypes, measures per-SNP and per-genotype accuracy against real data (and against simple PCA reconstructions of real training samples), and demonstrates the LD-based expansion on at least a chromosome-scale region.
  2. [Section 4.2, Table 3] The augmentation experiment is confounded by information leakage from the real training set. The diffusion model is trained on the full real training set, and the augmentation rows of Table 3 train classifiers on a small 'seed' subset (5%, 10%, 20%, 50%) plus synthetic samples generated from the full data. Since the generator has seen the 95% of real samples that are withheld from the seed subset, the accuracy gains could be due to the synthetic data reproducing or interpolating those withheld samples, rather than to the general utility of synthetic genotypes. This undermines the conclusion that synthetic data 'rescues' small-data settings. The authors should either retrain the generator on only the seed subset, or use a held-out real-data split for generating synthetic data, and then re-evaluate the augmentation gains.
  3. [Appendix, 'Discussion on Human Genotypes' and Section 5] The central novelty claim includes expansion of synthetic genotypes into full-length DNA-level genomes, but the expansion procedure is only described qualitatively: non-SNP polymorphic sites are filled in using 'the by LD principles statistically most likely variant,' and non-polymorphic sites are filled with the reference letter. No experiment evaluates the accuracy, haplotype consistency, or population-genetic plausibility of such expanded genomes. Since the paper explicitly distinguishes itself from prior work by claiming full-length genome generation, this step needs a concrete demonstration, for example by expanding a set of synthetic genotypes on a chromosome, comparing allele frequencies and LD patterns to real data, and reporting switch error or imputation accuracy against a real reference panel.
minor comments (4)
  1. [Section 3.2, Fig. 1] Figure 1 states 'PCA to 8 Dimensions' with '<1% compression Error,' but the text says the number of principal components per gene ranges between 1 and 8 and shorter vectors are zero-padded. This discrepancy should be clarified in the figure caption or text so that the fixed dimension is understood as a padded width, not the actual number of retained PCs for every gene.
  2. [Section 3.4.2, Eq. (3)] The notation in Eq. (3) is difficult to parse, especially the subscripts in 'PrivacyLoss = AAtruthtr,syn − AAtruthte,syn.' Please define each subscript (tr, te, syn, truth) explicitly and state over which pairs the nearest-neighbour comparisons are computed.
  3. [Section 4.2, Table 4] The 1KG MLP+CNN row reports a negative Privacy Loss (-0.2225), while the text states that values closer to 0 are better. A negative value of meaningful magnitude should be discussed, since it implies that synthetic samples are closer to test data than to training data in the adversarial metric, which is unusual and may indicate a different form of distribution mismatch.
  4. [Throughout] There are several typographical and formatting errors: the title contains 'DIFFU - SION MODELS,' Section 4.2 has 'ALs data' instead of 'ALS data,' and the appendix contains 'homozygoous' instead of 'homozygous.' A careful proofread is recommended.

Circularity Check

1 steps flagged · score 4.0 of 10

The classifier benchmark is self-contained, but the 'complete synthetic human genotypes' claim rests on a self-cited, unvalidated PCA-invertibility step.

  1. self citation load bearing [Section 3.2 (Embedding Genotypes / Haplotypes), with the claim reused in the Abstract and Section 5]
    "For both ALS and 1KG, the reduction of dimension comes at minimal compression loss ( < 1%) as shown in Luo et al. (2023). This strongly implies that one can decompress the PC embedded data into the original genotypes or haplotypes at only a minor loss of information."

    All generation and evaluation occur in the per-gene PCA embedding space (R^{18432x8} / R^{26624x8}); no synthetic embedding is ever inverted back to a SNP genotype or DNA sequence. The paper's headline claim that these are 'complete synthetic human genotypes' therefore depends entirely on the asserted decompressibility of the PCA embedding. That decompressibility is imported from Luo et al. (2023), a prior paper by overlapping authors (Luo and Schönhuth), and the <1% loss was measured on real data, not on generated samples. The Appendix's expansion to DNA-level genomes is likewise asserted via 'LD principles', not demonstrated. Thus the central genotype-level claim reduces to a self-citation rather than to any derivation in this paper.

full rationale

The main experimental evaluation—training classifiers on synthetic embeddings and testing them on held-out real embeddings (Tables 2, 3, 6)—is not circular: the test data are real and held out, and the recovery-rate metric is a legitimate comparison against a real-data-trained upper bound. The augmentation experiment in Table 3 has a leakage confound because the generator was trained on the full real training set before being used to augment small subsets, but that is a data-contamination/soundness issue, not equation-level circularity. The genuine circularity concern is narrower: the paper's central claim of generating 'complete synthetic human genotypes' that expand to 'full-length, DNA-level genomes' rests on the per-gene PCA embedding being invertible, and the only support offered is a self-cited result from Luo et al. (2023) with overlapping authorship. Because the <1% reconstruction loss was established on real data and never validated on synthetic samples, and because no synthetic embedding is ever decoded, the genotype-level claim is carried by self-citation rather than by any derivation in this paper. This is load-bearing, but the classification results still provide independent content, so the score is 4 rather than higher.

Assumptions & free parameters 1 free parameters · 4 assumptions · 0 invented entities

The paper introduces no new physical or biological entities. Its load-bearing external inputs are the PCA embedding fidelity assumption and the assumption that synthetic data does not statistically leak the real training samples; both are domain assumptions inherited from prior work or from the evaluation protocol, and neither is independently verified here.

free parameters (1)
  • Per-gene PCA component count = 1 to 8 (padded to 8)
    For each of 18,279 (ALS) or 26,624 (1KG) genes, the number of principal components retained is a hand-set compression level chosen to keep reconstruction error below 1%, per the embedding protocol of Luo et al. (2023). This determines how much genotype information reaches the diffusion model and is not justified independently of downstream performance.
assumptions (4)
  • domain assumption PCA embeddings with 1-8 PCs per gene preserve the genetic signal needed for downstream disease/population classification (compression error <1%, cited from Luo et al. 2023).
    Invoked in Sec. 3.2; the entire pipeline operates in embedding space, so if the embedding discards disease-relevant variation, the synthetic data will not transfer to genotype space.
  • domain assumption Genotypes can be expanded into full-length DNA-level genomes by inserting LD-inferred variants at non-SNP sites and reference letters elsewhere.
    The abstract claims expandability to DNA-level genomes; this assumes linkage disequilibrium makes non-SNP positions inferable, and does not account for structural variants, indels, or recombination; the expansion is never executed or validated (Appendix, 'DISCUSSION ON HUMAN GENOTYPES').
  • domain assumption The diffusion model does not memorize training genotypes, so synthetic data used for augmentation does not leak identifiable information from the real training samples.
    Invoked in Sec. 3.4.2 and 4.2; the NNAA checks only nearest-neighbor copying, not statistical leakage that could improve classifiers or reveal individuals; the ethics statement acknowledges this limitation.
  • standard math Standard diffusion process equations (Ho et al. 2020, Eqs. 4-9) correctly model the embedding distribution.
    Background mathematical framework, unproblematic.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Generating Synthetic Genotypes using Diffusion Models." pith.science (2026). https://pith.science/paper/EALESCFR

@misc{pith2026241203278,
  author       = {Pith},
  title        = {Pith review of: Generating Synthetic Genotypes using Diffusion Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EALESCFR}},
  note         = {Machine review of arXiv:2412.03278}
}
read the original abstract

In this paper, we introduce the first diffusion model designed to generate complete synthetic human genotypes, which, by standard protocols, one can straightforwardly expand into full-length, DNA-level genomes. The synthetic genotypes mimic real human genotypes without just reproducing known genotypes, in terms of approved metrics. When training biomedically relevant classifiers with synthetic genotypes, accuracy is near-identical to the accuracy achieved when training classifiers with real data. We further demonstrate that augmenting small amounts of real with synthetically generated genotypes drastically improves performance rates. This addresses a significant challenge in translational human genetics: real human genotypes, although emerging in large volumes from genome wide association studies, are sensitive private data, which limits their public availability. Therefore, the integration of additional, insensitive data when striving for rapid sharing of biomedical knowledge of public interest appears imperative.

Figures

Figures reproduced from arXiv: 2412.03278 by the authors.

Figure 1
Figure 1. Pre-processing pipeline For that transformation, we consider the genes recorded for the ALS and 1KG datasets, amounting to 18279 and 26624 genes, respec￾tively. Based on approved principles, we assign each SNP site to one of the genes. The num￾ber of SNP sites per gene can vary quite sub￾stantially. Depending on length and location in the genome, a gene can collect roughly between 5 and 100 SNP sites. In other words… view at source ↗
Figure 2
Figure 2. A structural overview of the architecture of [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. We display validation loss (a) and reconstruction error (b) e.g. [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: UMAP visualizations of data of different origins. Orange and grey points are test and train [PITH_FULL_IMAGE:figures/full_fig_p015_4.png]
Figure 5
Figure 5. Figure 5: Reconstruction error vs noise curves during training for different diffusion model back [PITH_FULL_IMAGE:figures/full_fig_p015_5.png]
Figure 6
Figure 6. Figure 6: On the left, λ vs t (noise amplitude) curves after training for the MLP + CNN diffusion model backbones. Higher lambda results in more weight on the CNN part of the network. On the right, λ averaged over noise levels on y-axis and training step on x-axis during trainin…
Figure 7
Figure 7. Figure 7: On the left, a TSNE dimension reduction of the bottleneck of the MLP diffusion model of [PITH_FULL_IMAGE:figures/full_fig_p016_7.png]
Figure 8
Figure 8. Figure 8: An overview of the CNN diffusion model architecture. [PITH_FULL_IMAGE:figures/full_fig_p017_8.png]
Figure 9
Figure 9. Figure 9: An overview of the transformer diffusion model architecture. [PITH_FULL_IMAGE:figures/full_fig_p018_9.png]
Figure 10
Figure 10. Figure 10: The (unphased) genotype counts the number of alternative alleles (blue) in the two an￾cestral genome copies each individual inherits. Note that expanding the genotype G for an in￾dividual into a full-length genome over the al￾phabet {A, C, G, T} corresponds to a strai…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

44 extracted references · 24 canonical work pages

  1. [1]

    European Journal of Human Genetics, 26 0 (10): 0 1537--1546, 2018

    Project mine: study design and pilot analyses of a large-scale whole-genome sequencing study in amyotrophic lateral sclerosis. European Journal of Human Genetics, 26 0 (10): 0 1537--1546, 2018

    work page 2018

  2. [2]

    Genome-ac-gan: Enhancing synthetic genotype generation through auxiliary classification

    Shaked Ahronoviz and Ilan Gronau. Genome-ac-gan: Enhancing synthetic genotype generation through auxiliary classification. bioRxiv, pp.\ 2024--02, 2024

    work page 2024

  3. [3]

    Imputation of exome sequence variants into population-based samples and blood-cell-trait-associated loci in african americans: Nhlbi go exome sequencing project

    Paul L Auer, Jill M Johnsen, Andrew D Johnson, Benjamin A Logsdon, Leslie A Lange, Michael A Nalls, Guosheng Zhang, Nora Franceschini, Keolu Fox, Ethan M Lange, et al. Imputation of exome sequence variants into population-based samples and blood-cell-trait-associated loci in african americans: Nhlbi go exome sequencing project. The American Journal of Hum...

    work page 2012

  4. [4]

    Dirichlet diffusion score model for biological sequence generation

    Pavel Avdeyev, Chenlai Shi, Yuhao Tan, Kseniia Dudnyk, and Jian Zhou. Dirichlet diffusion score model for biological sequence generation. In International Conference on Machine Learning, pp.\ 1276--1301. PMLR, 2023

    work page 2023

  5. [5]

    Shekoofeh Azizi, Simon Kornblith, Chitwan Saharia, Mohammad Norouzi, and David J. Fleet. Synthetic data from diffusion models improves imagenet classification. Transactions on Machine Learning Research, 2023. ISSN 2835-8856. URL https://openreview.net/forum?id=DlRsoxjyPm

    work page 2023

  6. [6]

    Generating realistic artificial human genomes using adversarial autoencoders

    Callum Burnard, Alban Mancheron, and William J Ritchie. Generating realistic artificial human genomes using adversarial autoencoders. bioRxiv, pp.\ 2023--12, 2023

    work page 2023

  7. [7]

    Accurate proteome-wide missense variant effect prediction with alphamissense

    Jun Cheng, Guido Novati, Joshua Pan, Clare Bycroft, Akvil \.e Z emgulyt \.e , Taylor Applebaum, Alexander Pritzel, Lai Hong Wong, Michal Zielinski, Tobias Sargeant, et al. Accurate proteome-wide missense variant effect prediction with alphamissense. Science, 381 0 (6664): 0 eadg7492, 2023

    work page 2023

  8. [8]

    A global reference for human genetic variation

    1000 Genomes Project Consortium et al. A global reference for human genetic variation. Nature, 526 0 (7571): 0 68, 2015

    work page 2015

Show all 44 references
  1. [9]

    Tractable and expressive generative models of genetic variation data

    Meihua Dang, Anji Liu, Xinzhu Wei, Sriram Sankararaman, and Guy Van den Broeck. Tractable and expressive generative models of genetic variation data. bioRxiv, 2023

    2023

  2. [10]

    BERT: pre-training of deep bidirectional transformers for language understanding

    Jacob Devlin, Ming - Wei Chang, Kenton Lee, and Kristina Toutanova. BERT: pre-training of deep bidirectional transformers for language understanding. In Jill Burstein, Christy Doran, and Thamar Solorio (eds.), Proceedings of the 2019 Conference of the North American Chapter of...

    2019

  3. [11]

    Diffusion models beat gans on image synthesis

    Prafulla Dhariwal and Alexander Nichol. Diffusion models beat gans on image synthesis. Advances in neural information processing systems, 34: 0 8780--8794, 2021

    2021

  4. [12]

    Detection of long repeat expansions from pcr-free whole-genome sequence data

    Egor Dolzhenko, Joke Jfa Van Vugt, Richard J Shaw, Mitchell A Bekritsky, Marka Van Blitterswijk, Giuseppe Narzisi, Subramanian S Ajay, Vani Rajan, Bryan R Lajoie, Nathan H Johnson, et al. Detection of long repeat expansions from pcr-free whole-genome sequence data. Genome rese...

    1903

  5. [13]

    An image is worth 16x16 words: Transformers for image recognition at scale

    Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, et al. An image is worth 16x16 words: Transformers for image recognition at scale. arXiv preprint arXiv...

    2010 arXiv

  6. [14]

    Diffusion models in bioinformatics and computational biology

    Zhiye Guo, Jian Liu, Yanli Wang, Mengrui Chen, Duolin Wang, Dong Xu, and Jianlin Cheng. Diffusion models in bioinformatics and computational biology. Nat Rev Bioeng, 2 0 (2): 0 136--154, October 2023

    2023

  7. [15]

    Gans trained by a two time-scale update rule converge to a local nash equilibrium

    Martin Heusel, Hubert Ramsauer, Thomas Unterthiner, Bernhard Nessler, and Sepp Hochreiter. Gans trained by a two time-scale update rule converge to a local nash equilibrium. Advances in neural information processing systems, 30, 2017

    2017

  8. [16]

    Classifier-free diffusion guidance

    Jonathan Ho and Tim Salimans. Classifier-free diffusion guidance. arXiv preprint arXiv:2207.12598, 2022

    2022 arXiv

  9. [17]

    Denoising diffusion probabilistic models

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising diffusion probabilistic models. 2020

    2020

  10. [18]

    Highly accurate protein structure prediction with alphafold

    John Jumper, Richard Evans, Alexander Pritzel, Tim Green, Michael Figurnov, Olaf Ronneberger, Kathryn Tunyasuvunakool, Russ Bates, Augustin Z \' dek, Anna Potapenko, et al. Highly accurate protein structure prediction with alphafold. Nature, 596 0 (7873): 0 583--589, 2021

    2021

  11. [19]

    Discdiff: Latent diffusion model for dna sequence generation

    Zehui Li, Yuhao Ni, William AV Beardall, Guoxuan Xia, Akashaditya Das, Guy-Bart Stan, and Yiren Zhao. Discdiff: Latent diffusion model for dna sequence generation. arXiv preprint arXiv:2402.06079, 2024

    2024 arXiv

  12. [20]

    Predicting the prevalence of complex genetic diseases from individual genotype profiles using capsule networks

    Xiao Luo, Xiongbin Kang, and Alexander Schönhuth. Predicting the prevalence of complex genetic diseases from individual genotype profiles using capsule networks. In Nature Machine Intelligence. Nature Publishing, 2023. ISBN 978-3-319-24574-4

    2023

  13. [21]

    Umap: Uniform manifold approximation and projection

    Leland McInnes, John Healy, Nathaniel Saul, and Lukas Großberger. Umap: Uniform manifold approximation and projection. Journal of Open Source Software, 3 0 (29): 0 861, 2018. doi:10.21105/joss.00861. URL https://doi.org/10.21105/joss.00861

    2018 doi

  14. [22]

    Baccus, and Chris Ré

    Eric Nguyen, Michael Poli, Marjan Faizi, Armin Thomas, Callum Birch-Sykes, Michael Wornow, Aman Patel, Clayton Rabideau, Stefano Massaroli, Yoshua Bengio, Stefano Ermon, Stephen A. Baccus, and Chris Ré. Hyenadna: Long-range genomic sequence modeling at single nucleotide resolu...

    2023

  15. [23]

    Generative moment matching networks for genotype simulation

    Maria Perera, Daniel Mas Montserrat, M \' riam Barrab \'e s, Margarita Geleta, Xavier Gir \'o -i Nieto, and Alexander G Ioannidis. Generative moment matching networks for genotype simulation. In 2022 44th Annual International Conference of the IEEE Engineering in Medicine & Bi...

    2022

  16. [24]

    High-resolution image synthesis with latent diffusion models

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Bj \"o rn Ommer. High-resolution image synthesis with latent diffusion models. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp.\ 10684--10695, 2022

    2022

  17. [25]

    U-net: Convolutional networks for biomedical image segmentation

    Olaf Ronneberger, Philipp Fischer, and Thomas Brox. U-net: Convolutional networks for biomedical image segmentation. In Nassir Navab, Joachim Hornegger, William M. Wells, and Alejandro F. Frangi (eds.), Medical Image Computing and Computer-Assisted Intervention -- MICCAI 2015,...

    2015

  18. [26]

    Improved techniques for training gans

    Tim Salimans, Ian Goodfellow, Wojciech Zaremba, Vicki Cheung, Alec Radford, and Xi Chen. Improved techniques for training gans. Advances in neural information processing systems, 29, 2016

    2016

  19. [27]

    Designing dna with tunable regulatory activity using discrete diffusion

    Anirban Sarkar, Ziqi Tang, Chris Zhao, and Peter Koo. Designing dna with tunable regulatory activity using discrete diffusion. bioRxiv, pp.\ 2024--05, 2024

    2024

  20. [28]

    Caduceus: Bi-directional equivariant long-range dna sequence modeling

    Yair Schiff, Chia-Hsiang Kao, Aaron Gokaslan, Tri Dao, Albert Gu, and Volodymyr Kuleshov. Caduceus: Bi-directional equivariant long-range dna sequence modeling. arXiv preprint arXiv:2403.03234, 2024

    2024 arXiv

  21. [29]

    Dna-diffusion: Leveraging generative models for controlling chromatin accessibility and gene expression via synthetic regulatory elements

    Simon Senan, Aniketh Janardhan Reddy, Zach Nussbaum, Aaron Wenteler, Matei Bejan, Michael I Love, Wouter Meuleman, and Luca Pinello. Dna-diffusion: Leveraging generative models for controlling chromatin accessibility and gene expression via synthetic regulatory elements. In IC...

    2024

  22. [31]

    Deep unsupervised learning using nonequilibrium thermodynamics

    Jascha Sohl-Dickstein, Eric Weiss, Niru Maheswaranathan, and Surya Ganguli. Deep unsupervised learning using nonequilibrium thermodynamics. In International conference on machine learning, pp.\ 2256--2265. PMLR, 2015

    2015

  23. [32]

    Understanding and mitigating copying in diffusion models

    Gowthami Somepalli, Vasu Singla, Micah Goldblum, Jonas Geiping, and Tom Goldstein. Understanding and mitigating copying in diffusion models. Advances in Neural Information Processing Systems, 36: 0 47783--47803, 2023

    2023

  24. [33]

    Denoising diffusion implicit models, 2022

    Jiaming Song, Chenlin Meng, and Stefano Ermon. Denoising diffusion implicit models, 2022

    2022

  25. [34]

    Towards creating longer genetic sequences with gans: Generation in principal component space

    Antoine Szatkownik, Cyril Furtlehner, Guillaume Charpiat, Burak Yelmen, and Flora Jay. Towards creating longer genetic sequences with gans: Generation in principal component space. In Machine Learning in Computational Biology, pp.\ 110--122. PMLR, 2024

    2024

  26. [35]

    Visualizing data using t-sne

    Laurens van der Maaten and Geoffrey Hinton. Visualizing data using t-sne. Journal of Machine Learning Research, 9 0 (86): 0 2579--2605, 2008. URL http://jmlr.org/papers/v9/vandermaaten08a.html

    2008

  27. [36]

    HAPNEST: efficient, large-scale generation and evaluation of synthetic datasets for genotypes and phenotypes

    Sophie Wharrie, Zhiyu Yang, Vishnu Raj, Remo Monti, Rahul Gupta, Ying Wang, Alicia Martin, Luke J O’Connor, Samuel Kaski, Pekka Marttinen, Pier Francesco Palamara, Christoph Lippert, and Andrea Ganna. HAPNEST: efficient, large-scale generation and evaluation of synthetic datas...

    2023 doi

  28. [37]

    Discovery of a structural class of antibiotics with explainable deep learning

    Felix Wong, Erica J Zheng, Jacqueline A Valeri, Nina M Donghia, Melis N Anahtar, Satotaka Omori, Alicia Li, Andres Cubillos-Ruiz, Aarti Krishnan, Wengong Jin, et al. Discovery of a structural class of antibiotics with explainable deep learning. Nature, 626 0 (7997): 0 177--185, 2024

    2024

  29. [38]

    Privacy preserving synthetic health data

    Andrew Yale, Saloni Dash, Ritik Dutta, Isabelle Guyon, Adrien Pavao, and Kristin P Bennett. Privacy preserving synthetic health data. In ESANN 2019-European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning, 2019

    2019

  30. [39]

    Creating artificial human genomes using generative neural networks

    Burak Yelmen, Aur \'e lien Decelle, Linda Ongaro, Davide Marnetto, Corentin Tallec, Francesco Montinaro, Cyril Furtlehner, Luca Pagani, and Flora Jay. Creating artificial human genomes using generative neural networks. PLoS genetics, 17 0 (2): 0 e1009303, 2021

    2021

  31. [40]

    Deep convolutional and conditional neural networks for large-scale genomic data generation

    Burak Yelmen, Aur \'e lien Decelle, Leila Lea Boulos, Antoine Szatkownik, Cyril Furtlehner, Guillaume Charpiat, and Flora Jay. Deep convolutional and conditional neural networks for large-scale genomic data generation. PLoS Computational Biology, 19 0 (10): 0 e1011584, 2023

    2023

  32. [41]

    Dnagpt: a generalized pretrained tool for multiple dna sequence analysis tasks

    Daoan Zhang, Weitong Zhang, Bing He, Jianguo Zhang, Chenchen Qin, and Jianhua Yao. Dnagpt: a generalized pretrained tool for multiple dna sequence analysis tasks. bioRxiv, pp.\ 2023--07, 2023

    2023

  33. [42]

    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 gl...

  34. [43]

    @esa (Ref

    \@ifxundefined[1] #1\@undefined \@firstoftwo \@secondoftwo \@ifnum[1] #1 \@firstoftwo \@secondoftwo \@ifx[1] #1 \@firstoftwo \@secondoftwo [2] @ #1 \@temptokena #2 #1 @ \@temptokena \@ifclassloaded agu2001 natbib The agu2001 class already includes natbib coding, so you should ...

  35. [44]

    \@lbibitem[] @bibitem@first@sw\@secondoftwo \@lbibitem[#1]#2 \@extra@b@citeb \@ifundefined br@#2\@extra@b@citeb \@namedef br@#2 \@nameuse br@#2\@extra@b@citeb \@ifundefined b@#2\@extra@b@citeb @num @parse #2 @tmp #1 NAT@b@open@#2 NAT@b@shut@#2 \@ifnum @merge>\@ne @bibitem@firs...

  36. [45]

    oup-authoring-template.cls

    @open @close @open @close and [1] URL: #1 \@ifundefined chapter * \@mkboth \@ifxundefined @sectionbib * \@mkboth * \@mkboth\@gobbletwo \@ifclassloaded amsart * \@ifclassloaded amsbook * \@ifxundefined @heading @heading NAT@ctr thebibliography [1] @ \@biblabel @NAT@ctr \@bibset...

    2022

Pith tools

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