pith. machine review for the scientific record. sign in

arxiv: 2605.02954 · v1 · submitted 2026-05-02 · 🧬 q-bio.GN · cs.LG

Recognition: unknown

EFGPP: Exploratory framework for genotype-phenotype prediction

Muhammad Muneeb , David B. Ascher
Authors on Pith no claims yet

Pith reviewed 2026-05-10 15:12 UTC · model grok-4.3

classification 🧬 q-bio.GN cs.LG
keywords genotype-phenotype predictionpolygenic risk scoresmigraineUK Biobankdata integrationAUCdepression GWAS
0
0 comments X

The pith

Combining multiple genetic and clinical data types improves migraine prediction accuracy over any single type alone.

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

The paper presents EFGPP as a framework for generating, ranking, and combining genotype-derived features, principal components, clinical and metabolomic covariates, and polygenic risk scores to predict complex traits. It tests this on migraine using UK Biobank data from 733 individuals, drawing polygenic scores from both migraine and depression genome-wide association studies. The central result is that the strongest single data source reached a test AUC of 0.644 while multi-source models reached 0.688 with migraine-focused inputs and 0.663 with depression-derived inputs. A reader would care because signals for complex diseases are scattered across data types, and a practical method for integrating them could make genetic prediction more reliable without requiring ever-larger single datasets.

Core claim

EFGPP generates genotype-derived features via PLINK, constructs polygenic risk scores with PRSice-2, AnnoPred, and LDAK-GWAS from migraine and depression GWAS, and merges these with principal components plus clinical and metabolomic covariates. On 733 UK Biobank migraine cases and controls, the best single data type produced a test AUC of 0.644; combining sources raised performance to 0.688 using migraine-focused inputs and to 0.663 using cross-trait depression inputs. Genetic features alone failed to beat the covariates-only baseline, yet genotype-derived features outperformed PRS alone and depression-derived PRS carried transferable signal.

What carries the argument

The EFGPP framework that generates, ranks, and integrates genotype-derived features, principal components, covariates, and polygenic risk scores from multiple GWAS sources for phenotype prediction.

If this is right

  • Genetic features alone do not outperform a covariates-only baseline in this migraine prediction setting.
  • Genotype-derived features outperform polygenic risk scores used in isolation.
  • Polygenic risk scores derived from a depression GWAS carry measurable predictive value for migraine.
  • The framework supplies a reproducible workflow for prioritising which data sources to combine for any given trait.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same ranking-and-combination steps could be applied to larger biobanks or to other traits where sample sizes allow direct comparison of single versus multi-source models.
  • Cross-trait transfer from depression to migraine suggests that scores trained on related phenotypes may systematically improve prediction for neurologically overlapping conditions.
  • If the ranking step reliably identifies the most informative data types, future studies could use it to decide which additional assays or cohorts are worth collecting.

Load-bearing premise

The modest gains from combining data types will continue to appear outside this migraine task, this sample of 733 individuals, and the particular covariates and polygenic score tools chosen.

What would settle it

Re-running the EFGPP pipeline on an independent cohort for a different complex trait and finding that no combination of data types exceeds the AUC of the single best data type would falsify the central claim.

Figures

Figures reproduced from arXiv: 2605.02954 by David B. Ascher, Muhammad Muneeb.

Figure 1
Figure 1. Figure 1: Overview of the EFGPP workflow. The framework proceeds through six main stages. (1) Data sources: heterogeneous inputs are assembled from the target cohort and external discovery resources, including genotype data, covariates, principal components, func￾tional annotations, and GWAS summary statistics. (2) Dataset generation: multiple individual datasets are constructed, including genotype-derived, covariat… view at source ↗
Figure 2
Figure 2. Figure 2: Fold-wise evaluation design used in EFGPP. The full dataset was partitioned into five stratified folds while preserving the migraine case–control ratio. For each fold, training, val￾idation, and held-out test subsets were defined. Training and validation subsets were used for dataset construction, model fitting, hyperparameter comparison, ranking, pruning, and multi￾modal combination selection. The held-ou… view at source ↗
Figure 3
Figure 3. Figure 3: Impact of data-generation parameters on training, validation, and test perfor￾mance for Configuration 1. (A) Best performance for each GWAS. (B) Comparative perfor￾mance across dataset types, shown using violin plots of AUC scores for each dataset category. (C) Impact of weight-file incorporation on model performance. (D) Performance by number of SNPs. (E) Comparison of PRS model performance. (F) Stability… view at source ↗
Figure 4
Figure 4. Figure 4: Impact of data-generation parameters on training, validation, and test perfor￾mance for Configuration 2. (A) Best performance across GWAS sources. (B) Comparative performance across dataset types. (C) Impact of weight-file incorporation on model performance. (D) Performance by number of SNPs. (E) Comparison of PRS model performance. (F) Stability analysis of machine-learning models across all datasets. Ana… view at source ↗
read the original abstract

Predicting complex human traits from genetic data is challenging because different genetic, clinical, and molecular data sources often contain different parts of the signal. Here, we present EFGPP, a reproducible framework for generating, ranking, and combining multiple types of data for genotype-to-phenotype prediction. We applied EFGPP to migraine prediction using UK Biobank data from 733 individuals. The framework combined genotype-derived features, principal components, clinical and metabolomic covariates, and polygenic risk scores generated from migraine and depression GWAS using PLINK, PRSice-2, AnnoPred, and LDAK-GWAS. The best single data type achieved a test AUC of 0.644, while combining multiple data types improved performance to 0.688 using migraine-focused inputs and 0.663 using cross-trait depression-derived inputs. Genetic features alone did not outperform the covariates-only baseline, but genotype-derived features performed better than PRS alone, and depression-derived PRS showed useful predictive signal. Overall, EFGPP provides a practical proof-of-concept framework for prioritising and integrating heterogeneous genetic data sources for complex phenotype prediction.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The manuscript presents EFGPP, a reproducible framework for generating, ranking, and combining genotype-derived features, principal components, clinical/metabolomic covariates, and polygenic risk scores (PRS) computed via multiple tools (PLINK, PRSice-2, AnnoPred, LDAK-GWAS) from migraine and depression GWAS summary statistics. Applied to migraine prediction in a UK Biobank sample of 733 individuals, the best single data type achieves a test AUC of 0.644 while multi-type integration reaches 0.688 (migraine-focused inputs) or 0.663 (cross-trait depression inputs); genetic features alone do not beat the covariates baseline, but the framework is positioned as a practical proof-of-concept for prioritizing heterogeneous data sources in complex-trait prediction.

Significance. If the modest AUC gains survive rigorous validation, EFGPP could offer a useful template for systematic integration of multi-source genetic and molecular data in prediction tasks. The explicit use of several PRS tools and cross-trait inputs is a constructive strength that highlights differential signal contributions; the open, reproducible design further supports its potential utility as an exploratory scaffold.

major comments (3)
  1. [Abstract] Abstract and Results: The reported test-AUC improvement (0.644 to 0.688) is presented without confidence intervals, a p-value on the delta, or any description of the cross-validation scheme, feature-ranking procedure, multiple-testing correction, or class-imbalance handling. With n=733 the test partition is necessarily small; absent these controls the 0.044 gain cannot be distinguished from sampling noise or overfitting.
  2. [Results] Methods/Results: No bootstrap, repeated random-split, or external-cohort results are supplied to test robustness of the combined-model superiority. The central claim that multi-type integration yields a practically useful lift therefore rests on a single, unreplicated point estimate whose stability is unknown.
  3. [Abstract] Abstract: The evaluation relies entirely on external GWAS summary statistics and off-the-shelf PRS software; no internal equations or cross-validation procedures are given that would reduce the reported AUCs to quantities fitted and tested strictly within the 733-individual cohort, leaving open the possibility that performance differences partly reflect external data leakage or tool-specific biases.
minor comments (2)
  1. [Abstract] Abstract: The sample size (n=733) and exact phenotype definition should be stated in the opening sentence for immediate context.
  2. [Results] The manuscript would benefit from a table summarizing the individual data-type AUCs and the exact feature counts entering each model.

Simulated Author's Rebuttal

3 responses · 1 unresolved

We thank the referee for their thorough review and constructive suggestions. We agree that the current presentation of results lacks sufficient statistical detail and robustness checks, which we will address in a revised version of the manuscript. Our responses to the major comments are as follows.

read point-by-point responses

  1. Referee: [Abstract] Abstract and Results: The reported test-AUC improvement (0.644 to 0.688) is presented without confidence intervals, a p-value on the delta, or any description of the cross-validation scheme, feature-ranking procedure, multiple-testing correction, or class-imbalance handling. With n=733 the test partition is necessarily small; absent these controls the 0.044 gain cannot be distinguished from sampling noise or overfitting.

    Authors: We acknowledge that the lack of confidence intervals and a formal test for the improvement makes it difficult to assess whether the gain is statistically meaningful. In the revised manuscript, we will include bootstrap-derived 95% confidence intervals for all reported AUC values and conduct a permutation-based test to evaluate the significance of the AUC difference between the best single data type and the integrated model. We will explicitly describe the cross-validation scheme in the Methods section, ensuring that all feature ranking and model fitting are performed exclusively on the training portion to prevent information leakage. Class imbalance is inherent in the migraine phenotype; we will add details on how it was handled, noting that AUC was used as the primary metric because it is robust to imbalance. We will also clarify that no multiple-testing correction was applied to the feature ranking step as it serves an exploratory purpose, and discuss this as a limitation. revision: yes

  2. Referee: [Results] Methods/Results: No bootstrap, repeated random-split, or external-cohort results are supplied to test robustness of the combined-model superiority. The central claim that multi-type integration yields a practically useful lift therefore rests on a single, unreplicated point estimate whose stability is unknown.

    Authors: We agree that demonstrating robustness beyond a single split would strengthen the findings. As this study is intended as an exploratory proof-of-concept rather than a definitive validation, we opted for a straightforward single-split evaluation. In the revision, we will perform and report results from 10 independent random train-test splits, providing mean AUC and standard deviation for the key models to illustrate the stability of the observed improvement. revision: yes

  3. Referee: [Abstract] Abstract: The evaluation relies entirely on external GWAS summary statistics and off-the-shelf PRS software; no internal equations or cross-validation procedures are given that would reduce the reported AUCs to quantities fitted and tested strictly within the 733-individual cohort, leaving open the possibility that performance differences partly reflect external data leakage or tool-specific biases.

    Authors: The framework is explicitly designed to leverage external GWAS summary statistics for PRS computation, which is the standard approach in the field to avoid overfitting to the target cohort. No individual-level phenotype data from the 733 UK Biobank samples was used in the generation of the PRS scores or the GWAS summary statistics. We will add explicit statements in the Methods and Discussion to clarify this and to address potential tool-specific biases by noting that we compared multiple PRS methods and selected based on performance within the training set. The AUCs reflect out-of-sample prediction on the held-out test set using these externally derived scores combined with internal covariates. revision: yes

standing simulated objections not resolved
  • We do not have access to an independent external cohort for additional validation.

Circularity Check

0 steps flagged

No significant circularity in derivation chain

full rationale

The paper describes an empirical exploratory framework (EFGPP) that applies standard, externally developed tools (PLINK, PRSice-2, AnnoPred, LDAK) to generate features from UK Biobank genotypes, external GWAS summary statistics, principal components, and covariates, then ranks and combines them via conventional machine-learning pipelines to produce held-out test AUCs. No equations, self-definitions, or fitted parameters are presented that reduce the reported performance numbers to quantities defined by construction within the same dataset. The central results are therefore independent of any internal circular reduction and rest on reproducible external inputs and standard statistical procedures.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The framework rests on standard domain assumptions of GWAS and PRS validity plus conventional statistical modeling practices; no new free parameters, axioms, or invented entities are introduced beyond those implicit in the cited software tools.

axioms (1)
  • domain assumption GWAS summary statistics and PRS calculation tools produce valid predictive signals for the target trait.
    Abstract relies on PRS generated from migraine and depression GWAS using PLINK, PRSice-2, AnnoPred, and LDAK-GWAS.

pith-pipeline@v0.9.0 · 5494 in / 1353 out tokens · 49416 ms · 2026-05-10T15:12:12.437406+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

39 extracted references · 38 canonical work pages

  1. [1]

    Visscher, P .M., Wray, N.R., Zhang, Q., Sklar, P ., McCarthy, M.I., Brown, M.A., and Y ang,607 J. (2017). 10 years of gwas discovery: Biology, function, and translation. The American608 Journal of Human Genetics101, 5–22. URL:http://dx.doi.org/10.1016/j.ajhg.2017.609 06.005. doi:10.1016/j.ajhg.2017.06.005. 610

    work page doi:10.1016/j.ajhg.2017.609 2017

  2. [2]

    (2022).611 Human genotype-to-phenotype predictions: Boosting accuracy with nonlinear models.612 PLOS ONE17, e0273293

    Medvedev, A., Mishra Sharma, S., Tsatsorin, E., Nabieva, E., and Y arotsky, D. (2022).611 Human genotype-to-phenotype predictions: Boosting accuracy with nonlinear models.612 PLOS ONE17, e0273293. URL:http://dx.doi.org/10.1371/journal.pone.0273293.613 doi:10.1371/journal.pone.0273293. 614

    work page doi:10.1371/journal.pone.0273293.613 2022

  3. [3]

    Chalmer, M.A., Esserlind, A.L., Olesen, J., and Hansen, T.F . (2018). Polygenic risk score:615 use in migraine research. J Headache Pain19, 29–10. doi:10.1186/s10194-018-0856-0.616

    work page doi:10.1186/s10194-018-0856-0.616 2018

  4. [4]

    (2013).617 Personalized medicine: From genotypes and molecular phenotypes towards therapy- ses-618 sion introduction

    LISTGARTEN, J., STEGLE, O., MORRIS, Q., BRENNER, S.E., and PARTS, L. (2013).617 Personalized medicine: From genotypes and molecular phenotypes towards therapy- ses-618 sion introduction. In Biocomputing 2014. WORLD SCIENTIFIC pp. 224–228. URL:619 http://dx.doi.org/10.1142/9789814583220_0022. doi:10.1142/9789814583220_0022.620

    work page doi:10.1142/9789814583220_0022 2013

  5. [5]

    Dong, X., Xiao, T., Chen, B., Lu, Y ., and Zhou, W. (2022). Precision medicine via the621 integration of phenotype-genotype information in neonatal genome project. Fundamen-622 tal Research2, 873–884. URL:http://dx.doi.org/10.1016/j.fmre.2022.07.003. doi:623 10.1016/j.fmre.2022.07.003. 624

    work page doi:10.1016/j.fmre.2022.07.003 2022

  6. [6]

    Kogelman, L.J.A., Esserlind, A.L., Francke Christensen, A., Awasthi, S., Ripke, S., Inga-625 son, A., Davidsson, O.B., Erikstrup, C., Hjalgrim, H., Ullum, H., Olesen, J., and Folk-626 mann Hansen, T. (2019). Migraine polygenic risk score associates with efficacy of migraine-627 specific drugs. Neurol Genet5, e364–e364. doi:10.1212/NXG.0000000000000364. Go...

    work page doi:10.1212/nxg.0000000000000364 2019

  7. [7]

    Genetic risk score analysis indicates migraine with and without636 comorbid depression are genetically different disorders

    Ligthart, L., Hottenga, J.J., Lewis, C.M., Farmer, A.E., Craig, I.W., Breen, G., Willem-633 sen, G., Vink, J.M., Middeldorp, C.M., Byrne, E.M., Heath, A.C., Madden, P .A.F ., Perga-634 dia, M.L., Montgomery, G.W., Martin, N.G., Penninx, B.W.J.H., McGuffin, P ., Boomsma,635 D.I., and Nyholt (2014). Genetic risk score analysis indicates migraine with and wi...

    work page doi:10.1007/s00439-013-1370-8 2014

  8. [8]

    Muneeb, M., Feng, S., and Henschel, A. (2022). Transfer learning for genotype–phenotype639 prediction using deep learning models. BMC Bioinformatics23. URL:http://dx.doi.org/640 10.1186/s12859-022-05036-8. doi:10.1186/s12859-022-05036-8.641

    work page doi:10.1186/s12859-022-05036-8 2022

  9. [9]

    Hunter, D.J. (2005). Gene–environment interactions in human diseases. Nature Reviews642 Genetics6, 287–298. URL:http://dx.doi.org/10.1038/nrg1578. doi:10.1038/nrg1578.643

    work page doi:10.1038/nrg1578 2005

  10. [10]

    Guo, T., and Li, X. (2023). Machine learning for predicting phenotype from genotype and644 environment. Current Opinion in Biotechnology79, 102853. URL:http://dx.doi.org/10.645 1016/j.copbio.2022.102853. doi:10.1016/j.copbio.2022.102853.646 21

    work page doi:10.1016/j.copbio.2022.102853.646 2023

  11. [11]

    Fadista, J., Manning, A.K., Florez, J.C., and Groop, L. (2016). The (in)famous gwas p-647 value threshold revisited and updated for low-frequency variants. European Journal of648 Human Genetics24, 1202–1205. URL:http://dx.doi.org/10.1038/ejhg.2015.269. doi:649 10.1038/ejhg.2015.269. 650

    work page doi:10.1038/ejhg.2015.269 2016

  12. [12]

    McCaw, Z.R., Colthurst, T., Yun, T., Furlotte, N.A., Carroll, A., Alipanahi, B., McLean,651 C.Y ., and Hormozdiari, F . (2022). Deepnull models non-linear covariate effects to im-652 prove phenotypic prediction and association power. Nature Communications13. URL:653 http://dx.doi.org/10.1038/s41467-021-27930-0. doi:10.1038/s41467-021-27930-0.654

    work page doi:10.1038/s41467-021-27930-0 2022

  13. [13]

    Principal components analysis corrects for stratification in genome-wide association656 studies

    Price, A.L., Patterson, N.J., Plenge, R.M., Weinblatt, M.E., Shadick, N.A., and Reich, D.655 (2006). Principal components analysis corrects for stratification in genome-wide association656 studies. Nature Genetics38, 904–909. URL:http://dx.doi.org/10.1038/ng1847. doi:657 10.1038/ng1847. 658

    work page doi:10.1038/ng1847 2006

  14. [14]

    Torkamani, A., Scott-Van Zeeland, A.A., Topol, E.J., and Schork, N.J. (2011). Annotating659 individual human genomes. Genomics98, 233–241. URL:http://dx.doi.org/10.1016/660 j.ygeno.2011.07.006. doi:10.1016/j.ygeno.2011.07.006.661

    work page doi:10.1016/660 2011

  15. [15]

    Tam, V., Patel, N., Turcotte, M., Boss´e, Y ., Par´e, G., and Meyre, D. (2019). Benefits and lim-662 itations of genome-wide association studies. Nature Reviews Genetics20, 467–484. URL:663 http://dx.doi.org/10.1038/s41576-019-0127-1. doi:10.1038/s41576-019-0127-1.664

    work page doi:10.1038/s41576-019-0127-1 2019

  16. [16]

    Torkamani, A., Wineinger, N.E., and Topol, E.J. (2018). The personal and clinical utility of665 polygenic risk scores. Nature Reviews Genetics19, 581–590. URL:http://dx.doi.org/666 10.1038/s41576-018-0018-x. doi:10.1038/s41576-018-0018-x.667

    work page doi:10.1038/s41576-018-0018-x 2018

  17. [17]

    Khera, A.V., Chaffin, M., Aragam, K.G., Haas, M.E., Roselli, C., Choi, S.H., Natara-668 jan, P ., Lander, E.S., Lubitz, S.A., Ellinor, P .T., and Kathiresan, S. (2018). Genome-wide669 polygenic scores for common diseases identify individuals with risk equivalent to mono-670 genic mutations. Nature Genetics50, 1219–1224. URL:http://dx.doi.org/10.1038/671 s...

    work page doi:10.1038/671 2018

  18. [18]

    Sehrawat, S., Najafian, K., and Jin, L. (2023). Predicting phenotypes from novel genomic673 markers using deep learning. Bioinformatics Advances3. URL:http://dx.doi.org/10.674 1093/bioadv/vbad028. doi:10.1093/bioadv/vbad028. 675

    work page doi:10.1093/bioadv/vbad028 2023

  19. [19]

    Ritchie, M.D., Holzinger, E.R., Li, R., Pendergrass, S.A., and Kim, D. (2015). Methods of676 integrating data to uncover genotype–phenotype interactions. Nature Reviews Genetics16,677 85–97. URL:http://dx.doi.org/10.1038/nrg3868. doi:10.1038/nrg3868.678

    work page doi:10.1038/nrg3868 2015

  20. [20]

    Garreta, L., Cer ´on-Souza, I., Palacio, M.R., and Reyes-Herrera, P .H. (2021). Multigwas:679 An integrative tool for genome wide association studies in tetraploid organisms. Ecol-680 ogy and Evolution11, 7411–7426. URL:http://dx.doi.org/10.1002/ece3.7572. doi:681 10.1002/ece3.7572. 682

    work page doi:10.1002/ece3.7572 2021

  21. [21]

    Multi-trait analysis of genome-wide association summary statistics using mtag

    Turley, P ., Walters, R.K., Maghzian, O., Okbay, A., Lee, J.J., Fontana, M.A., Nguyen-683 Viet, T.A., Wedow, R., Zacher, M., Furlotte, N.A., Magnusson, P ., Oskarsson, S., Johan-684 nesson, M., Visscher, P .M., Laibson, D., Cesarini, D., Neale, B.M., and Benjamin, D.J.685 (2018). Multi-trait analysis of genome-wide association summary statistics using mta...

    work page doi:10.1038/s41588-017-0009-4 2018

  22. [22]

    Ishigaki, K., Sakaue, S., Terao, C., Luo, Y ., Sonehara, K., Y amaguchi, K., Amariuta, T.,689 Too, C.L., Laufer, V.A., Scott, I.C., Viatte, S., Takahashi, M., Ohmura, K., Murasawa, A.,690 Hashimoto, M., Ito, H., Hammoudeh, M., Emadi, S.A., Masri, B.K., Halabi, H., Badsha,691 H., Uthman, I.W., Wu, X., Lin, L., Li, T., Plant, D., Barton, A., Orozco, G., Ver...

    work page doi:10.1038/s41588-022-01213-w 2022

  23. [23]

    Truong, B., Hull, L.E., Ruan, Y ., Huang, Q.Q., Hornsby, W., Martin, H., van Heel, D.A.,705 Wang, Y ., Martin, A.R., Lee, S.H., and Natarajan, P . (2024). Integrative polygenic risk706 score improves the prediction accuracy of complex traits and diseases. Cell Genomics4,707 100523. URL:http://dx.doi.org/10.1016/j.xgen.2024.100523. doi:10.1016/j.xgen.708 2...

    work page doi:10.1016/j.xgen.2024.100523 2024

  24. [24]

    Purcell, S., Neale, B., Todd-Brown, K., Thomas, L., Ferreira, M.A., Bender, D., Maller,710 J., Sklar, P ., de Bakker, P .I., Daly, M.J., and Sham, P .C. (2007). Plink: A tool set for711 whole-genome association and population-based linkage analyses. The American Jour-712 nal of Human Genetics81, 559 ˆaC“575. URL:http://dx.doi.org/10.1086/519795. doi:713 1...

    work page doi:10.1086/519795 2007

  25. [25]

    Choi, S.W., and O ˆaC™Reilly, P .F . (2019). Prsice-2: Polygenic risk score software for715 biobank-scale data. GigaScience8. URL:http://dx.doi.org/10.1093/gigascience/716 giz082. doi:10.1093/gigascience/giz082. 717

    work page doi:10.1093/gigascience/716 2019

  26. [26]

    Zheng, Z., Liu, S., Sidorenko, J., Wang, Y ., Lin, T., Y engo, L., Turley, P ., Ani, A., Wang,718 R., Nolte, I.M., Snieder, H., Aguirre-Gamboa, R., Deelen, P ., Franke, L., Kuivenhoven, J.A.,719 Lopera Maya, E.A., Sanna, S., Swertz, M.A., Vonk, J.M., Wijmenga, C., Y ang, J., Wray,720 N.R., Goddard, M.E., Visscher, P .M., and Zeng, J. (2024). Leveraging fu...

    work page doi:10.1038/s41588-024-01704-y.724 2024

  27. [27]

    Zhang, Q., Priv ´e, F ., Vilhj ´almsson, B., and Speed, D. (2021). Improved genetic pre-725 diction of complex traits from individual-level data or summary statistics. Nature Com-726 munications12. URL:http://dx.doi.org/10.1038/s41467-021-24485-y. doi:10.1038/727 s41467-021-24485-y. 728

    work page doi:10.1038/s41467-021-24485-y 2021

  28. [28]

    Ligthart, L., Hottenga, J.J., Lewis, C.M., Farmer, A.E., Craig, I.W., Breen, G., Willemsen,729 G., Vink, J.M., Middeldorp, C.M., Byrne, E.M., Heath, A.C., Madden, P .A.F ., Pergadia, M.L.,730 Montgomery, G.W., Martin, N.G., Penninx, B.W.J.H., McGuffin, P ., Boomsma, D.I., and Ny-731 holt, D.R. (2014). Genetic risk score analysis indicates migraine with an...

    2014

  29. [29]

    Muneeb, M., and Ascher, D.B. (2026). A harmonized benchmarking framework for734 implementation-aware evaluation of 46 polygenic risk score tools across binary and continu-735 ous phenotypes. URL:http://dx.doi.org/10.64898/2026.03.22.713457. doi:10.64898/736 2026.03.22.713457. 737

    work page doi:10.64898/2026.03.22.713457 2026

  30. [30]

    Common genetic associations between age-related diseases

    D ¨onertas ¸, H.M., Fabian, D.K., Fuentealba, M., Partridge, L., and Thornton, J.M.738 (2021). Common genetic associations between age-related diseases. Nature Ag-739 ing1, 400–412. URL:http://dx.doi.org/10.1038/s43587-021-00051-5. doi:10.1038/740 s43587-021-00051-5. 741

    work page doi:10.1038/s43587-021-00051-5 2021

  31. [31]

    Jiang, L., Zheng, Z., Fang, H., and Y ang, J. (2021). A generalized linear mixed model742 association tool for biobank-scale data. Nature Genetics53, 1616–1621. URL:http://dx.743 doi.org/10.1038/s41588-021-00954-4. doi:10.1038/s41588-021-00954-4.744

    work page doi:10.1038/s41588-021-00954-4 2021

  32. [32]

    Wray, N.R., Ripke, S., Mattheisen, M., Trzaskowski, M., Byrne, E.M., Abdellaoui, A., Adams,745 M.J., Agerbo, E., Air, T.M., Andlauer, T.M.F ., Bacanu, S.A., Bækvad-Hansen, M., Beek-746 man, A.F .T., Bigdeli, T.B., Binder, E.B., Blackwood, D.R.H., Bryois, J., Buttenschøn, H.N.,747 Bybjerg-Grauholm, J., Cai, N., Castelao, E., Christensen, J.H., Clarke, T.K....

    work page doi:10.1038/s41588-018-0090-3 2018

  33. [33]

    Harder, A., Nguyen, T.D., Pasman, J.A., Mosing, M.A., H¨agg, S., and Lu, Y . (2022). Genetics763 of age-at-onset in major depression. Translational Psychiatry12. URL:http://dx.doi.764 org/10.1038/s41398-022-01888-z. doi:10.1038/s41398-022-01888-z.765

    work page doi:10.1038/s41398-022-01888-z 2022

  34. [34]

    Quality control procedures for genome-wide association studies

    Turner, S., Armstrong, L.L., Bradford, Y ., Carlson, C.S., Crawford, D.C., Crenshaw, A.T.,766 de Andrade, M., Doheny, K.F ., Haines, J.L., Hayes, G., Jarvik, G., Jiang, L., Kullo, I.J.,767 Li, R., Ling, H., Manolio, T.A., Matsumoto, M., McCarty, C.A., McDavid, A.N., Mirel, D.B.,768 Paschall, J.E., Pugh, E.W., Rasmussen, L.V., Wilke, R.A., Zuvich, R.L., an...

    work page doi:10.1002/0471142905.hg0119s68.771 2011

  35. [35]

    Truong, V.Q., Woerner, J.A., Cherlin, T.A., Bradford, Y ., Lucas, A.M., Okeh, C.C., Shiv-773 akumar, M.K., Hui, D.H., Kumar, R., Pividori, M., Jones, S.C., Bossa, A.C., Turner, S.D.,774 Ritchie, M.D., and Verma, S.S. (2022). Quality control procedures for genome-wide as-775 sociation studies. Current Protocols2. URL:http://dx.doi.org/10.1002/cpz1.603. doi...

    work page doi:10.1002/cpz1.603 2022

  36. [36]

    Anderson, C.A., Pettersson, F .H., Clarke, G.M., Cardon, L.R., Morris, A.P ., and Zonder-778 van, K.T. (2010). Data quality control in genetic case-control association studies. Na-779 ture Protocols5, 1564–1573. URL:http://dx.doi.org/10.1038/nprot.2010.116. doi:780 10.1038/nprot.2010.116. 781

    work page doi:10.1038/nprot.2010.116 2010

  37. [37]

    (2020).782 Recommendations for choosing the genotyping method and best practices for quality783 control in crop genome-wide association studies

    Pavan, S., Delvento, C., Ricciardi, L., Lotti, C., Ciani, E., and D’Agostino, N. (2020).782 Recommendations for choosing the genotyping method and best practices for quality783 control in crop genome-wide association studies. Frontiers in Genetics11. URL:http:784 //dx.doi.org/10.3389/fgene.2020.00447. doi:10.3389/fgene.2020.00447.785

    work page doi:10.3389/fgene.2020.00447 2020

  38. [38]

    Zeng, J., Zheng, Z., Liu, S., Sidorenko, J., Y engo, L., Turley, P ., Ani, A., Wang, R.,786 Nolte, I., Snieder, H., Y ang, J., Wray, N., Goddard, M., and Visscher, P . (2023). Lever-787 aging functional genomic annotations and genome coverage to improve polygenic pre-788 diction of complex traits within and between ancestries. European Neuropsychopharma-7...

    work page doi:10.1016/j.euroneuro.2023.08.063 2023

  39. [39]

    Massey, F .J. (1951). The kolmogorov-smirnov test for goodness of fit. Journal of the Ameri-792 can Statistical Association46, 68–78. URL:http://dx.doi.org/10.1080/01621459.1951.793 10500769. doi:10.1080/01621459.1951.10500769. 794 25

    work page doi:10.1080/01621459.1951.793 1951