Pith. sign in

REVIEW 3 major objections 6 minor 49 references

PhenoKG: Knowledge Graph-Driven Gene Discovery and Patient Insights from Phenotypes Alone

T0 review · 3 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

Pith's one-line read PhenoKG predicts a patient's causative gene from phenotypes alone, reporting higher MRR than the previous best method on both real and simulated rare-disease cohorts.

desk verdict PhenoKG is a credible candidate-free gene prioritization method with strong simulated performance, but its real-data advantage over SHEPHERD is not statistically supported as reported. read the letter →

arxiv 2506.13119 v1 pith:ZQFNXJ6J submitted 2025-06-16 cs.LG cs.AIcs.NEq-bio.GNq-bio.QM

classification cs.LGcs.AIcs.NEq-bio.GNq-bio.QM
keywords geneprioritizationphenotype-drivendiagnosisknowledgegraphneuralnetworksrarediseasetransformerMyGene2PrimeKG
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

PhenoKG is a method for rare Mendelian disease diagnosis that takes only a patient's phenotype terms and produces a ranked list of likely causative genes, with no requirement for a pre-filtered candidate list or sequencing data. It builds a patient-specific subgraph from the PrimeKG knowledge graph, encodes it with graph attention layers and a transformer, and ranks roughly 8,000 genes by cosine similarity to the patient embedding. On the real MyGene2 cohort the authors report an MRR of 24.64% and nDCG@100 of 33.64%, above SHEPHERD's 19.02% and 30.54%; on the simulated UDN-like test set, the best phenotype-only configuration reaches 91.08% MRR versus 79.35% for SHEPHERD. The paper argues that such a phenotype-only model can act as an upstream triage tool or pre-filter before variant analysis, and that patient embeddings capture biological structure beyond the single causative gene.

What carries the argument

The load-bearing object is the patient-specific subgraph $G_p$: the nodes and edges on shortest paths, or in the $k$-hop neighborhood, connecting each patient's HPO phenotypes to candidate genes in PrimeKG. A GATv2 graph encoder refines node embeddings, a transformer encoder produces gene embeddings, and multi-head attention with learnable memory vectors pools the phenotype embeddings into a patient vector; the final relevance score is the cosine similarity between patient and gene vectors. Two losses carry the training: a semi-hard-negative triplet loss that pulls the true gene closer than the hardest plausible negative, and a patient-similarity loss, backed by a memory bank of past patients, that pulls together patients who share a causative gene. The subgraph simultaneously defines the candidate gene set and supplies the relational structure that the learned embeddings need.

What would settle it

Run PhenoKG and SHEPHERD on all 146 MyGene2 patients without the two-hop exclusion, or on an independent cohort where some verified causative genes are known to sit three or more hops from the patient's phenotype terms; if MRR drops toward or below SHEPHERD's 19.02% in that setting, the conditional advantage does not generalize.

Watch

Extended reading notes

Core claim

The central claim is that causative-gene prioritization can be solved from phenotypes alone by enriching HPO terms with a knowledge graph, and that the resulting model outperforms the leading baseline under identical evaluation conditions. PhenoKG reaches an MRR of 24.64% and nDCG@100 of 33.64% on the real MyGene2 dataset without a candidate gene list, compared with 19.02% and 30.54% for SHEPHERD, and its top-1 accuracy rises to 83.96% when an expert-curated candidate list of about 20 genes is available. The same architecture, trained with a combination of a gene-ranking triplet loss and a patient-similarity loss, is reported to reach 91.08% MRR on a simulated test set structured like the Undiagnosed Diseases Network. The authors frame the model as a pre-filter and prioritization step that narrows the search space for clinicians, not as a standalone diagnostic.

Load-bearing premise

The reported MyGene2 numbers assume that a patient's true causative gene lies within two hops of the patient's phenotypes in PrimeKG, because the 25 of 146 patients whose genes were unreachable were excluded from the evaluation.

Editorial extensions

If this is right

  • On MyGene2, the no-candidate-list setting reaches 24.64% MRR and 33.64% nDCG@100, compared with SHEPHERD's 19.02% and 30.54%.
  • With a candidate list of roughly 20 genes, top-1 accuracy is reported at 83.96%, so the model serves as both a discovery layer and a re-ranker.
  • On the simulated UDN-like test set, the best phenotype-only configuration reaches 91.08% MRR, above SHEPHERD's 79.35%.
  • The combined gene and patient-similarity loss is the most reliable configuration on the real dataset, indicating that aligning patients with the same causative gene supports generalization.
  • Because the model starts from HPO terms alone and ranks around 8,000 genes, it can operate as a pre-filter before whole-genome variant analysis rather than after it.

Reading between the lines

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

  • The evaluation on MyGene2 drops 25 of 146 patients whose true gene is not reachable within two hops, so the reported advantage is conditional; a full-cohort or multi-hop evaluation would likely lower the numbers, and that gap is the main thing prospective validation should measure.
  • The paper's observation that patient embeddings cluster beyond shared causative genes suggests a use it does not develop: phenotype-based patient matching or cohort discovery for undiagnosed patients.
  • A concrete next experiment would take genes ranked highly by PhenoKG that are absent from expert panels, subject them to variant testing, and measure how often an unsuspected gene is confirmed; the paper explicitly says this requires prospective validation.
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 / 6 minor

Summary. The paper introduces PhenoKG, a phenotype-only gene prioritization model that builds patient-specific subgraphs from the PrimeKG knowledge graph, encodes them with GATv2 layers and a transformer gene encoder, and ranks candidate genes via contrastive and patient-similarity losses. The model is trained on simulated rare-disease patients and evaluated on a simulated test set and on the real MyGene2 dataset. The central claim is that PhenoKG outperforms the SHEPHERD baseline: MRR of 24.64% versus 19.02% on MyGene2 and MRR of 91.08% versus 79.35% on the simulated test set, with additional ablations across loss functions and embedding initializations.

Significance. If the real-data comparison were sound, this would be a practically useful contribution: phenotype-only gene prioritization that does not require an expert-curated candidate list could serve as an upstream triage step in rare-disease diagnostics. The paper has genuine strengths: it uses an external real-world benchmark (MyGene2), trains on simulated data rather than on the test labels, reports multiple seeds for its own model, and provides a fairly detailed ablation of loss components and embedding initializations. However, the headline real-data result is not yet convincingly established because the evaluation protocol differs from the baseline in a way that can only favor PhenoKG, and the reported gap is not statistically supported.

major comments (3)
  1. [Section 4.1, Table 2] The MyGene2 evaluation is restricted to the 121 of 146 patients whose causative gene is reachable within a two-hop neighborhood in PrimeKG, and the candidate set is exactly the genes in that two-hop neighborhood. This means the reported MRR is conditional on a favorable reachability assumption, and for the 25 excluded patients the model has zero chance by construction. The comparison with SHEPHERD is therefore not apples-to-apples unless SHEPHERD is evaluated on the same 121 patients with the same candidate set, or PhenoKG is evaluated genome-wide. This issue is load-bearing for the abstract's claim of 'substantial improvements' on MyGene2.
  2. [Section 4.4, Table 2] The claimed MyGene2 improvement over SHEPHERD is not statistically established. The best PhenoKG configuration reports MRR 24.64 ± 4.57 across three seeds, while SHEPHERD is reported as 19.02 ± 0.00. The 5.62-point gap is only about 1.2 standard deviations of the PhenoKG estimate, and no paired per-patient comparison or significance test is provided. The authors should report paired metrics on the same patients, confidence intervals, or a permutation test before claiming superiority.
  3. [Section 4.4, Tables 1 and 2] The paper selects the best of nine PhenoKG configurations (three loss combinations times three embedding strategies) on each dataset and compares that single best configuration against one SHEPHERD number. No model-selection procedure on a validation set is described for the MyGene2 results, and no multiple-comparison correction is applied. The comparison should either specify a pre-registered or validation-based selection rule, or report all configurations collectively (e.g., with error bars that include selection variability).
minor comments (6)
  1. [Section 3.1] The sentence 'The set G with the goal of identifying the causative gene g∗ ∈ G' is grammatically incomplete, and the notation G is used both for the knowledge graph and for the candidate gene set, which is confusing.
  2. [Section 3.3] The regularization term is written as λ|∥p∥2 + mean(∥G∥2) − 2|, but the intended norm is not clear; this should be a squared L2 norm or the formula should be clarified, especially since the text says 'close to unit norm'.
  3. [Section 4.4] The text says 'nDCG@111.57%' and 'nDCG@115.15±3.33%' where the intended metric is nDCG@1; the repeated '1' is a typo that should be corrected.
  4. [Section 4.2] The phrase 'The patient neighbourhood was defined using k=2 nearest neighbours' is ambiguous: k is used earlier for the k-hop neighborhood in the knowledge graph, but here it appears to describe something else; please clarify the notation.
  5. [Figure 3] The x-axis labels are not fully legible and the figure caption does not specify the exact metric plotted ('Match percentage' is defined only implicitly); please add a clear description of the evaluation protocol for this figure.
  6. [Section 3.3] The memory bank M used in the patient similarity loss is described only as 'maintain[ing] a memory bank', with no details on its initialization, size, or update schedule beyond the circular buffer; please provide these details for reproducibility.

Circularity Check

1 steps flagged · score 6.0 of 10

MyGene2 headline MRR is scored in a KG-closed world: patient eligibility and the candidate set are the same two-hop PrimeKG reachability criterion, so the answer is always present by construction.

  1. self definitional [Section 4.1 (Dataset, MyGene2 filter) with Section 3.4 (Inference candidate set); headline claim in Abstract and Section 4.4/Table 2]
    "As MyGene2 lacks expert-curated candidate gene lists, we used it to test model performance without candidate genes. For efficiency, we restricted evaluation to the two-hop neighborhood around each patient; patients whose causative gene was unreachable within two hops were excluded, leaving 121 patients. ... At inference time, we construct the patient-specific subgraph G_p by extracting the k-hop neighborhood around each phenotype in P_p, thereby capturing the set of candidate genes G_p associated with the patient's clinical profile."

    Patient eligibility and the model's output space are defined by the same criterion: a patient is scored only if the true gene lies within the two-hop PrimeKG neighborhood of the phenotypes, and the model ranks only genes in that same neighborhood. The answer is thus guaranteed to be in the candidate set for every scored patient, and the 25 excluded patients, where the KG lacks the connection, have zero chance by definition. Since gene embeddings are pretrained via link prediction on that same PrimeKG (Section 3.1), the 24.64% MRR is a closed-world retrieval of associations the KG already encodes, conditional on the answer being KG-reachable.

full rationale

PhenoKG's architecture and simulated-data evaluation are self-contained: the model is trained on the Alsentzer et al. simulated cohort and tested on a held-out 320-patient split with reported seed variance, and the Table 1 margin over SHEPHERD (79.35 to 91.08 MRR) is an external benchmark comparison with independent content. There is no load-bearing self-citation; the authors cite Alsentzer et al. (2022, 2023) as prior external work with no author overlap, and no uniqueness theorem is imported from the authors' own prior papers. The partial circularity is confined to the headline real-data claim (Abstract, Section 4.4, Table 2). The MyGene2 protocol defines patient eligibility by two-hop reachability of the true gene in PrimeKG and defines the model's output space as that same two-hop neighborhood, so the correct gene is guaranteed to be in the candidate set for every scored patient, and the 25 excluded patients are precisely the cases where the model has zero chance by definition. Because the gene embeddings are pretrained on link prediction over the same PrimeKG, the reported 24.64% MRR is a closed-world retrieval score conditioned on the KG already encoding the answer's proximity; the Abstract's unqualified 'surpassing the best baseline (SHEPHERD) at 19.02% MRR' omits this conditioning, and the Conclusion's claim of finding 'novel or previously unobserved phenotype-gene associations not explicitly represented in the knowledge graph' is contradicted by the exclusion of unreachable genes. The 5.62-point gap versus a 4.57 seed-to-seed standard deviation, with SHEPHERD shown as 19.02±0.00, is a separate statistical robustness risk rather than an instance of circularity. Overall, the central real-data claim partially reduces, by construction of the filter and candidate set, to a KG-conditional closed-world statement; the underlying ranking and the simulated benchmark remain independent, so the paper is partially, not fully, circular.

Assumptions & free parameters 6 free parameters · 4 assumptions · 0 invented entities

No new biological entities, forces, or mediators are proposed. The model introduces learnable memory vectors and a memory bank of patient embeddings, which are computational constructs, not biological inventions. The main scientific assumptions are about the completeness and correctness of the underlying knowledge graph and the representativeness of simulated training data.

free parameters (6)
  • k (KG neighborhood radius) = 2
    Chosen by hand; defines the candidate gene set for patients without a candidate list. The causative gene must be within two hops of the phenotypes in PrimeKG, and patients outside this radius are excluded from MyGene2 evaluation. This parameter directly controls the achievable ceiling of the method.
  • margin gamma (triplet loss) = 0.3
    Chosen by hand in Section 4.2; controls the separation margin between positive and negative gene similarities in the triplet loss.
  • regularization weight lambda = 0.03
    Chosen by hand in Section 4.2; weights the L2 norm regularization term in the gene loss.
  • patient similarity temperature alpha = 0.5
    Chosen by hand in Section 4.2; scales the logits in the patient similarity pull loss.
  • patient similarity margin delta = 0.8
    Chosen by hand in Section 4.2; margin in the push term of the patient similarity loss.
  • contrastive temperature tau = 0.12 (initial, learnable)
    Initialized at 0.12 and described as a learnable temperature in Section 3.3; tuned during training.
assumptions (4)
  • domain assumption PrimeKG accurately reflects gene-phenotype-disease relationships relevant for rare disease gene prioritization.
    Section 4.1 describes using PrimeKG as the backbone; the method's candidate set and node embeddings derive from this KG. If the KG omits or misrepresents key associations, the model cannot rank the true gene.
  • domain assumption The simulated dataset from Alsentzer et al. 2023 is a valid training proxy for real patient phenotype data.
    Section 4.1 trains on simulated patients because real labeled data are scarce; the transfer to MyGene2 assumes the simulated phenotype distributions and candidate lists resemble clinical reality.
  • ad hoc to paper The causative gene, when it exists, is reachable within the k-hop neighborhood of the patient's phenotypes in the KG.
    Section 4.1: 'patients whose causative gene was unreachable within two hops were excluded, leaving 121 patients'. This is an evaluability assumption that filters the test set; it is not a property of all real patients.
  • domain assumption HPO phenotype terms assigned to patients are correct and complete for the causative gene's phenotype spectrum.
    Both training and test data rely on HPO annotations; incomplete or noisy phenotype terms would weaken the phenotype-gene signal.

how reviews work

0 comments
Cite this review

Pith. "Pith review of PhenoKG: Knowledge Graph-Driven Gene Discovery and Patient Insights from Phenotypes Alone." pith.science (2026). https://pith.science/paper/ZQFNXJ6J

@misc{pith2026250613119,
  author       = {Pith},
  title        = {Pith review of: PhenoKG: Knowledge Graph-Driven Gene Discovery and Patient Insights from Phenotypes Alone},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZQFNXJ6J}},
  note         = {Machine review of arXiv:2506.13119}
}
read the original abstract

Identifying causative genes from patient phenotypes remains a significant challenge in precision medicine, with important implications for the diagnosis and treatment of genetic disorders. We propose a novel graph-based approach for predicting causative genes from patient phenotypes, with or without an available list of candidate genes, by integrating a rare disease knowledge graph (KG). Our model, combining graph neural networks and transformers, achieves substantial improvements over the current state-of-the-art. On the real-world MyGene2 dataset, it attains a mean reciprocal rank (MRR) of 24.64\% and nDCG@100 of 33.64\%, surpassing the best baseline (SHEPHERD) at 19.02\% MRR and 30.54\% nDCG@100. We perform extensive ablation studies to validate the contribution of each model component. Notably, the approach generalizes to cases where only phenotypic data are available, addressing key challenges in clinical decision support when genomic information is incomplete.

Figures

Figures reproduced from arXiv: 2506.13119 by the authors.

Figure 1
Figure 1. Overview of PhenoKG for rare disease gene prioritization. The model constructs a patient-specific subgraph [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. An example of a simplified patient-specific one-hop subgraph and its neighborhood, illustrating the relation [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Match percentage comparison for MyGene2 (left) and simulated test (right) datasets, across top- [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

49 extracted references · 43 canonical work pages

  1. [1]

    Addressing diagnostic gaps and priorities of the global rare diseases community: Recommendations from the irdirc diagnostics scientific committee

    David R Adams, Clara DM van Karnebeek, Sergi Beltran Agullo, V \' ctor Faundes, Saumya Shekhar Jamuar, Sally Ann Lynch, Guillem Pintos-Morell, Ratna Dua Puri, Ruty Shai, Charles A Steward, et al. Addressing diagnostic gaps and priorities of the global rare diseases community: Recommendations from the irdirc diagnostics scientific committee. European Journ...

  2. [2]

    The ensembl gene annotation system

    Bronwen L Aken, Sarah Ayling, Daniel Barrell, Laura Clarke, Valery Curwen, Susan Fairley, Julio Fernandez Banet, Konstantinos Billis, Carlos Garc \' a Gir \'o n, Thibaut Hourlier, et al. The ensembl gene annotation system. Database, 2016: 0 baw093, 2016

  3. [3]

    Few shot learning for phenotype-driven diagnosis of patients with rare genetic diseases

    Emily Alsentzer, Michelle M Li, Shilpa N Kobren, Ayush Noori, Undiagnosed Diseases Network, Isaac S Kohane, and Marinka Zitnik. Few shot learning for phenotype-driven diagnosis of patients with rare genetic diseases. medRxiv, pages 2022--12, 2022

  4. [4]

    Simulation of undiagnosed patients with novel genetic conditions

    Emily Alsentzer, Samuel G Finlayson, Michelle M Li, Undiagnosed Diseases Network, Shilpa N Kobren, and Isaac S Kohane. Simulation of undiagnosed patients with novel genetic conditions. Nature Communications, 14 0 (1): 0 6403, 2023

  5. [5]

    Personalised analytics for rare disease diagnostics

    Denise Anderson, Gareth Baynam, Jenefer M Blackwell, and Timo Lassmann. Personalised analytics for rare disease diagnostics. Nature communications, 10 0 (1): 0 5274, 2019

  6. [6]

    Layer normalization

    Jimmy Lei Ba, Jamie Ryan Kiros, and Geoffrey E Hinton. Layer normalization. arXiv preprint arXiv:1607.06450, 2016

  7. [7]

    Amelie speeds mendelian diagnosis by matching patient phenotype and genotype to primary literature

    Johannes Birgmeier, Maximilian Haeussler, Cole A Deisseroth, Ethan H Steinberg, Karthik A Jagadeesh, Alexander J Ratner, Harendra Guturu, Aaron M Wenger, Mark E Diekhans, Peter D Stenson, et al. Amelie speeds mendelian diagnosis by matching patient phenotype and genotype to primary literature. Science Translational Medicine, 12 0 (544): 0 eaau9113, 2020

  8. [8]

    Rare-disease genetics in the era of next-generation sequencing: discovery to translation

    Kym M Boycott, Megan R Vanstone, Dennis E Bulman, and Alex E MacKenzie. Rare-disease genetics in the era of next-generation sequencing: discovery to translation. Nature Reviews Genetics, 14 0 (10): 0 681--691, 2013

Show all 49 references
  1. [9]

    How attentive are graph attention networks? In International Conference on Learning Representations, 2022

    Shaked Brody, Uri Alon, and Eran Yahav. How attentive are graph attention networks? In International Conference on Learning Representations, 2022. URL https://openreview.net/forum?id=F72ximsx7C1

  2. [10]

    Building a knowledge graph to enable precision medicine

    Payal Chandak, Kexin Huang, and Marinka Zitnik. Building a knowledge graph to enable precision medicine. Scientific Data, 10 0 (1): 0 67, 2023

  3. [11]

    The genetic basis of mendelian phenotypes: discoveries, challenges, and opportunities

    Jessica X Chong, Kati J Buckingham, Shalini N Jhangiani, Corinne Boehm, Nara Sobreira, Joshua D Smith, Tanya M Harrell, Margaret J McMillin, Wojciech Wiszniewski, Tomasz Gambin, et al. The genetic basis of mendelian phenotypes: discoveries, challenges, and opportunities. The A...

  4. [12]

    Opportunities and challenges for machine learning in rare diseases

    Sergio Decherchi, Elena Pedrini, Marina Mordenti, Andrea Cavalli, and Luca Sangiorgi. Opportunities and challenges for machine learning in rare diseases. Frontiers in medicine, 8: 0 747612, 2021

  5. [13]

    Neural networks for classification and image generation of aging in genetic syndromes

    Dat Duong, Ping Hu, Cedrik Tekendo-Ngongang, Suzanna E Ledgister Hanchard, Simon Liu, Benjamin D Solomon, and Rebekah L Waikel. Neural networks for classification and image generation of aging in genetic syndromes. Frontiers in genetics, 13: 0 864092, 2022

  6. [14]

    The national institutes of health undiagnosed diseases program: insights into rare diseases

    William A Gahl, Thomas C Markello, Camilo Toro, Karin Fuentes Fajardo, Murat Sincan, Fred Gill, Hannah Carlson-Donohoe, Andrea Gropman, Tyler Mark Pierson, Gretchen Golas, et al. The national institutes of health undiagnosed diseases program: insights into rare diseases. Genet...

  7. [15]

    Identifying facial phenotypes of genetic disorders using deep learning

    Yaron Gurovich, Yair Hanani, Omri Bar, Guy Nadav, Nicole Fleischer, Dekel Gelbman, Lina Basel-Salmon, Peter M Krawitz, Susanne B Kamphausen, Martin Zenker, et al. Identifying facial phenotypes of genetic disorders using deep learning. Nature medicine, 25 0 (1): 0 60--64, 2019

  8. [16]

    Genetic syndromes screening by facial recognition technology: Vgg-16 screening model construction and evaluation

    Dian Hong, Ying-Yi Zheng, Ying Xin, Ling Sun, Hang Yang, Min-Yin Lin, Cong Liu, Bo-Ning Li, Zhi-Wei Zhang, Jian Zhuang, et al. Genetic syndromes screening by facial recognition technology: Vgg-16 screening model construction and evaluation. Orphanet Journal of Rare Diseases, 1...

  9. [17]

    Pedia: prioritization of exome data by image analysis

    Tzung-Chien Hsieh, Martin A Mensah, Jean T Pantel, Dione Aguilar, Omri Bar, Allan Bayat, Luis Becerra-Solano, Heidi B Bentzen, Saskia Biskup, Oleg Borisov, et al. Pedia: prioritization of exome data by image analysis. Genetics in Medicine, 21 0 (12): 0 2807--2814, 2019

  10. [18]

    Gestaltmatcher facilitates rare disease matching using facial phenotype descriptors

    Tzung-Chien Hsieh, Aviram Bar-Haim, Shahida Moosa, Nadja Ehmke, Karen W Gripp, Jean Tori Pantel, Magdalena Danyel, Martin Atta Mensah, Denise Horn, Stanislav Rosnev, et al. Gestaltmatcher facilitates rare disease matching using facial phenotype descriptors. Nature genetics, 54...

  11. [19]

    M-cap eliminates a majority of variants of uncertain significance in clinical exomes at high sensitivity

    Karthik A Jagadeesh, Aaron M Wenger, Mark J Berger, Harendra Guturu, Peter D Stenson, David N Cooper, Jonathan A Bernstein, and Gill Bejerano. M-cap eliminates a majority of variants of uncertain significance in clinical exomes at high sensitivity. Nature genetics, 48 0 (12): ...

  12. [20]

    Phrank measures phenotype sets similarity to greatly improve mendelian diagnostic disease prioritization

    Karthik A Jagadeesh, Johannes Birgmeier, Harendra Guturu, Cole A Deisseroth, Aaron M Wenger, Jonathan A Bernstein, and Gill Bejerano. Phrank measures phenotype sets similarity to greatly improve mendelian diagnostic disease prioritization. Genetics in Medicine, 21 0 (2): 0 464...

  13. [21]

    A visual and curatorial approach to clinical variant prioritization and disease gene discovery in genome-wide diagnostics

    Regis A James, Ian M Campbell, Edward S Chen, Philip M Boone, Mitchell A Rao, Matthew N Bainbridge, James R Lupski, Yaping Yang, Christine M Eng, Jennifer E Posey, et al. A visual and curatorial approach to clinical variant prioritization and disease gene discovery in genome-w...

  14. [22]

    a rvelin and Jaana Kek \

    Kalervo J \"a rvelin and Jaana Kek \"a l \"a inen. Cumulated gain-based evaluation of ir techniques. ACM Transactions on Information Systems (TOIS), 20 0 (4): 0 422--446, 2002

  15. [23]

    Phen-gen: combining phenotype and genotype to analyze rare disorders

    Asif Javed, Saloni Agrawal, and Pauline C Ng. Phen-gen: combining phenotype and genotype to analyze rare disorders. Nature methods, 11 0 (9): 0 935--937, 2014

  16. [24]

    o hler, Marcel H Schulz, Peter Krawitz, Sebastian Bauer, Sandra D \

    Sebastian K \"o hler, Marcel H Schulz, Peter Krawitz, Sebastian Bauer, Sandra D \"o lken, Claus E Ott, Christine Mundlos, Denise Horn, Stefan Mundlos, and Peter N Robinson. Clinical diagnostics in human genetics with semantic similarity searches in ontologies. The American Jou...

  17. [25]

    Expansion of the human phenotype ontology (hpo) knowledge base and resources

    Sebastian K \"o hler, Leigh Carmody, Nicole Vasilevsky, Julius O B Jacobsen, Daniel Danis, Jean-Philippe Gourdine, Michael Gargano, Nomi L Harris, Nicolas Matentzoglu, Julie A McMurry, et al. Expansion of the human phenotype ontology (hpo) knowledge base and resources. Nucleic...

  18. [26]

    Imagenet classification with deep convolutional neural networks

    Alex Krizhevsky, Ilya Sutskever, and Geoffrey E Hinton. Imagenet classification with deep convolutional neural networks. Advances in neural information processing systems, 25, 2012

  19. [27]

    Xrare: a machine learning method jointly modeling phenotypes and genetic evidence for rare disease diagnosis

    Qigang Li, Keyan Zhao, Carlos D Bustamante, Xin Ma, and Wing H Wong. Xrare: a machine learning method jointly modeling phenotypes and genetic evidence for rare disease diagnosis. Genetics in Medicine, 21 0 (9): 0 2126--2134, 2019

  20. [28]

    Sgdr: Stochastic gradient descent with warm restarts

    Ilya Loshchilov and Frank Hutter. Sgdr: Stochastic gradient descent with warm restarts. arXiv preprint arXiv:1608.03983, 2016

  21. [29]

    Decoupled weight decay regularization

    Ilya Loshchilov and Frank Hutter. Decoupled weight decay regularization. arXiv preprint arXiv:1711.05101, 2017

  22. [30]

    Rectifier nonlinearities improve neural network acoustic models

    Andrew L Maas, Awni Y Hannun, Andrew Y Ng, et al. Rectifier nonlinearities improve neural network acoustic models. In Proc. icml, volume 30, page 3. Atlanta, GA, 2013

  23. [31]

    Ai-marrvel—a knowledge-driven ai system for diagnosing mendelian disorders

    Dongxue Mao, Chaozhong Liu, Linhua Wang, Rami AI-Ouran, Cole Deisseroth, Sasidhar Pasupuleti, Seon Young Kim, Lucian Li, Jill A Rosenfeld, Linyan Meng, et al. Ai-marrvel—a knowledge-driven ai system for diagnosing mendelian disorders. NEJM AI, 1 0 (5): 0 AIoa2300009, 2024

  24. [32]

    Estimating cumulative point prevalence of rare diseases: analysis of the orphanet database

    St \'e phanie Nguengang Wakap, Deborah M Lambert, Annie Olry, Charlotte Rodwell, Charlotte Gueydan, Val \'e rie Lanneau, Daniel Murphy, Yann Le Cam, and Ana Rath. Estimating cumulative point prevalence of rare diseases: analysis of the orphanet database. European journal of hu...

  25. [33]

    Cada: phenotype-driven gene prioritization based on a case-enriched knowledge graph

    Chengyao Peng, Simon Dieck, Alexander Schmid, Ashar Ahmad, Alexej Knaus, Maren Wenzel, Laura Mehnert, Birgit Zirn, Tobias Haack, Stephan Ossowski, et al. Cada: phenotype-driven gene prioritization based on a case-enriched knowledge graph. NAR Genomics and Bioinformatics, 3 0 (...

  26. [34]

    Insights into genetics, human biology and disease gleaned from family based genomic studies

    Jennifer E Posey, Anne H O’Donnell-Luria, Jessica X Chong, Tamar Harel, Shalini N Jhangiani, Zeynep H Coban Akdemir, Steven Buyske, Davut Pehlivan, Claudia MB Carvalho, Samantha Baxter, et al. Insights into genetics, human biology and disease gleaned from family based genomic ...

  27. [35]

    Rarebert: transformer architecture for rare disease patient identification using administrative claims

    PKS Prakash, Srinivas Chilukuri, Nikhil Ranade, and Shankar Viswanathan. Rarebert: transformer architecture for rare disease patient identification using administrative claims. In Proceedings of the AAAI conference on artificial intelligence, volume 35, pages 453--460, 2021

  28. [36]

    The undiagnosed diseases network: accelerating discovery about health and disease

    Rachel B Ramoni, John J Mulvihill, David R Adams, Patrick Allard, Euan A Ashley, Jonathan A Bernstein, William A Gahl, Rizwan Hamid, Joseph Loscalzo, Alexa T McCray, et al. The undiagnosed diseases network: accelerating discovery about health and disease. The American Journal ...

  29. [37]

    Phenotype-driven gene prioritization for rare diseases using graph convolution on heterogeneous networks

    Aditya Rao, Saipradeep Vg, Thomas Joseph, Sujatha Kotte, Naveen Sivadasan, and Rajgopal Srinivasan. Phenotype-driven gene prioritization for rare diseases using graph convolution on heterogeneous networks. BMC medical genomics, 11: 0 1--12, 2018

  30. [38]

    Cadd: predicting the deleteriousness of variants throughout the human genome

    Philipp Rentzsch, Daniela Witten, Gregory M Cooper, Jay Shendure, and Martin Kircher. Cadd: predicting the deleteriousness of variants throughout the human genome. Nucleic acids research, 47 0 (D1): 0 D886--D894, 2019

  31. [39]

    Sue Richards, Nazneen Aziz, Sherri Bale, David Bick, Soma Das, Julie Gastier-Foster, Wayne W Grody, Madhuri Hegde, Elaine Lyon, Elaine Spector, et al. Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the american college...

  32. [40]

    Interpretable clinical genomics with a likelihood ratio paradigm

    Peter N Robinson, Vida Ravanmehr, Julius OB Jacobsen, Daniel Danis, Xingmin Aaron Zhang, Leigh C Carmody, Michael A Gargano, Courtney L Thaxton, Guy Karlebach, Justin Reese, et al. Interpretable clinical genomics with a likelihood ratio paradigm. The American Journal of Human ...

  33. [41]

    A deep learning frame-work for recognizing developmental disorders

    Pushkar Shukla, Tanu Gupta, Aradhya Saini, Priyanka Singh, and Raman Balasubramanian. A deep learning frame-work for recognizing developmental disorders. In 2017 IEEE Winter Conference on Applications of Computer Vision (WACV), pages 705--714. IEEE, 2017

  34. [42]

    a ger, Sebastian K \

    Damian Smedley, Julius OB Jacobsen, Marten J \"a ger, Sebastian K \"o hler, Manuel Holtgrewe, Max Schubach, Enrico Siragusa, Tomasz Zemojtel, Orion J Buske, Nicole L Washington, et al. Next-generation diagnostics and disease-gene discovery with the exomiser. Nature protocols, ...

  35. [43]

    Mutationtaster2021

    Robin Steinhaus, Sebastian Proft, Markus Schuelke, David N Cooper, Jana Marie Schwarz, and Dominik Seelow. Mutationtaster2021. Nucleic Acids Research, 49 0 (W1): 0 W446--W451, 2021

  36. [44]

    Evaluating the clinical validity of gene-disease associations: an evidence-based framework developed by the clinical genome resource

    Natasha T Strande, Erin Rooney Riggs, Adam H Buchanan, Ozge Ceyhan-Birsoy, Marina DiStefano, Selina S Dwight, Jenny Goldstein, Rajarshi Ghosh, Bryce A Seifert, Tam P Sneddon, et al. Evaluating the clinical validity of gene-disease associations: an evidence-based framework deve...

  37. [45]

    Large language models with retrieval-augmented generation for zero-shot disease phenotyping

    Will E Thompson, David M Vidmar, Jessica K De Freitas, John M Pfeifer, Brandon K Fornwalt, Ruijun Chen, Gabriel Altay, Kabir Manghnani, Andrew C Nelsen, Kellie Morland, et al. Large language models with retrieval-augmented generation for zero-shot disease phenotyping. arXiv pr...

  38. [46]

    University of Washington, Center for Mendelian Genomics . Mygene2. https://www.mygene2.org/MyGene2/. Accessed: 2025-05-12

  39. [47]

    Attention is all you need

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, ukasz Kaiser, and Illia Polosukhin. Attention is all you need. Advances in neural information processing systems, 30, 2017

  40. [48]

    Phenolyzer: phenotype-based prioritization of candidate genes for human diseases

    Hui Yang, Peter N Robinson, and Kai Wang. Phenolyzer: phenotype-based prioritization of candidate genes for human diseases. Nature methods, 12 0 (9): 0 841--843, 2015

  41. [49]

    Inphernet accelerates monogenic disease diagnosis using patients’ candidate genes’ neighbors

    Boyoung Yoo, Johannes Birgmeier, Jonathan A Bernstein, and Gill Bejerano. Inphernet accelerates monogenic disease diagnosis using patients’ candidate genes’ neighbors. Genetics in Medicine, 23 0 (10): 0 1984--1992, 2021

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.