A Multi-Omic “Digital Embryo” Framework to Model Early Human Preimplantation Development In Silico

VIDEO TRANSCRIPT

One in six couples worldwide face infertility. And for most, in vitro fertilization, or IVF, is their best hope. But nearly 50% of IVF embryos arrest before becoming viable pregnancies, and nobody knows why.

Scientists have studied embryos one omic layer at a time, often missing the cross-omic interactions that could explain arrest.

Now, the Digital Embryo is the first framework to integrate six omics layers across 1,963 cells into one computational atlas for studying embryology. This multi-omic integration yielded novel biological discoveries regarding embryo arrest that no single-omic analysis could have identified.

In healthcare, a doctor could simulate diseases in a patient’s embryo, screen thousands of treatments, and identify the best intervention, all without any invasive tests. The framework could even be adapted to other animals to combat the extinction of endangered species.

But all the science aside, what I hope this does is help every couple struggling to conceive have the best possible chance. Too many of these potentially beautiful stories may have ended. But conception is, by its very nature, all about new beginnings. So let's hope the Digital Embryo brings many, many more.

The Problem

Every year, roughly 2.5 million IVF cycles are performed worldwide, and approximately 50% of the embryos created in those cycles arrest before they can ever become a viable pregnancy. In most cases, nobody can tell you why. Not the doctor. Not the clinic. Nobody.

One in six couples worldwide experience infertility. For them, every failed cycle means thousands of dollars, weeks of hormone injections, and a phone call that ends with “I’m sorry, you won’t have a baby this time.” The cruelest part isn’t the cost; it’s wondering what happened.

The Gap

Here’s what makes this problem so striking: we can sequence a person’s entire genome for $200, but the embryologist choosing which embryo gets a chance at life is still essentially grading it by how it looks under a microscope. The molecular state, or omics, of the embryo (gene expression, protein activity, metabolic health, epigenetic markers) is currently invisible to the clinicians making the decision.

Scientists have studied these omic layers, but almost always one at a time. That means we’re looking at a complex biological puzzle with only one piece visible, missing the cross-layer interactions that could possibly explain why embryos fail.

The Question

I asked: What if we could integrate six types of omic data into a single computational framework to predict which embryos are at risk, reveal why they fail, and even screen treatments before trying them in a clinic?

That question became The Digital Embryo.

1. Data Curation

I collected 9 datasets totaling 1,963 individual embryo cells across 6 omic layers: transcriptomics, proteomics, metabolomics, secretomics, genomics, and epigenomics.

2. Constructing the Transcriptomic Backbone

The 3 transcriptomic datasets were processed through a standardized pipeline: quality trimming, alignment to the T2T-CHM13v2.0 reference genome, duplicate marking, and gene counting. I used scVI to correct batch effects and produce a 16-dimensional latent space.

3. Layering the Other Omics

The transcriptomic backbone provides single-cell resolution, but the remaining five omic layers were measured at stage-level resolution. To address this, each layer’s measurements were encoded as features assigned to cells by their developmental stage. The result is a 94-dimensional feature matrix spanning all 6 omics.

4. Prediction

I trained an XGBoost classifier on the feature matrix to predict whether an embryo would arrest or continue developing. I validated it with leave-one-study-out cross-validation.

5. Explainability and Biological Analysis

I applied SHAP to assign each of the 94 features an importance score per cell, revealing which measurements most influence the model’s decision-making process. I inferred a gene regulatory network to map which transcription factors control which downstream genes. I also performed cross-omic concordance analysis to assess whether findings in one omic layer are supported by another.

6. The Perturbation Engine

The engine simulates the effect of a compound, culture condition, or environmental exposure by modifying an embryo's 94-dimensional multi-omic profile and re-predicting arrest probability through the XGBoost classifier.

I also designed a combinatorial treatment-discovery algorithm that generates pairwise and triplet combinations of protective compounds to formulate novel multi-omics treatments against arrest.

This engine extends to disease simulation: a disease’s molecular signature is applied to an embryo’s profile, candidate treatments are simulated, and the engine recalculates arrest probability for each, identifying the most effective treatments.

The Atlas and Trajectory

All 1,963 cells resolve into a single continuous trajectory from oocyte to late blastocyst. The 3 transcriptomic datasets intermix cleanly after batch correction. RNA velocity and CellRank recovered the expected developmental ordering without being told the stage labels, identifying 4 terminal states.

Prediction and Explainability

On unseen datasets, the XGBoost classifier achieves 0.870 AUC versus 0.754 for the transcriptomics-only control classifier—a +0.116 AUC gain from multi-omic integration. Ablation tests confirmed that removing any single omic decreases performance, with transcriptomics causing the largest drop. SHAP explainability validates this: transcriptomics ranks highest at 35.6% of feature importance. The remaining five omics each contribute between 5% and 18%.

Arrest is Structured, Not Random

Differential expression analysis identified 3,680 genes differentially expressed between arrested and normal embryos, with 720 to 1,182 per developmental stage. At the 8-cell stage, ZGA markers like TPRX2 and PRAMEF25 fail to properly activate, signaling that the embryo’s own genome never fully turned on. By the blastocyst stage, the failure shifts to metabolism: metabolomic features become 3.1 times more predictively important. Underlying both stages, gene regulatory network inference identified 15 regulatory edges among 10 transcription factors, including master pluripotency regulators and lineage specification factors.

Cross-Omic Convergence

When independent omic layers—analyzed separately, with no shared statistical pipeline—converge on the same biological conclusion, it constitutes strong evidence that the finding is real.

Protein-RNA Concordance

Protein fold changes matched transcript fold changes for 17/17 genes tested (p<0.00001).

The DNA Damage Response Hypermethylation Discovery

On chromosome 17, BRCA1 promoter methylation rises from 32% to 56% in arrested embryos. On chromosome 6, TP53 methylation climbs from 23% to 46%. Simultaneously, these same DNA damage response genes are transcriptionally downregulated. The epigenome silences the exact repair genes that expression data independently flags as underactive. This coordinated epigenomic-transcriptomic shutdown of DNA damage response has not been reported as a mechanism of preimplantation arrest in any published study.

Secretomic Enrichment

Genes targeted by embryo-secreted miRNAs were 12.3 times enriched among differentially expressed arrest genes (p=3x10-16), suggesting embryos may communicate molecular distress through their secretome.

Screening 85 Compounds

The perturbation engine screened 85 compounds by simulating their molecular effects across all six omics. Every protective compound reduced the predicted arrest probability, and every harmful compound increased it, indicating the engine has 100% directional accuracy. Of the 40 strongest predicted effects, 37 matched known experimental results from published in vitro studies.

Novel Treatment Combinations

The Digital Embryo generated 35 novel treatment combinations against arrest. The top pair, GM-CSF + Low O2 at 5%, and the top triplet, LIF + Low O2 at 5% + Resveratrol, both achieve coverage across all 6 omic layers. The triplet predicts a 75% reduction in arrest probability.

Disease Simulation

The Digital Embryo currently offers 10 curated disease profiles, each enabling clinicians to simulate diseases in a patient’s specific embryo, test compounds, adjust culture conditions, and identify the best treatment strategy before applying it in real life.

A New Way to Study Embryos

By integrating six omics, the Digital Embryo has revealed what single-omic analyses cannot: arrest is not random; it is a structured, multi-omic event. This is evidenced by unprecedented cross-omic signatures in DNA damage response genes and 35 novel multi-omic treatment combinations against arrest. These findings provide proof of concept that a multi-omic approach can fundamentally enrich our understanding of embryo biology, opening avenues of research that would not have been otherwise possible.

The Digital Embryo Taught Itself What Researchers Already Knew

The perturbation engine had never been shown any clinical data. Yet when it screened 85 compounds, 7/20 treatments were already standard IVF practice, and another 8 were active areas of published research. The model, using only computational resources, independently reached conclusions that took decades of laboratory experimentation to establish. This points toward a future in which reproductive medicine can be dramatically accelerated at a fraction of the cost.

From Trial-and-Error to Personalized Medicine

Today, when an IVF embryo arrests, clinicians adjust culture conditions based on general guidelines and hope the next cycle goes better. The Digital Embryo offers a different, non-invasive path by computationally screening thousands of treatments to identify the one most likely to help a specific patient’s embryo. Approximately 43% of arrested embryos were predicted to be rescuable with targeted treatment combinations, suggesting that nearly half of the embryos currently discarded could be saved. For the one in six couples worldwide facing infertility, that difference means everything.

More Data, Higher Resolution

I plan to incorporate additional datasets and pursue true single-cell resolution across all omics to improve model precision.

A More Representative Atlas

Available embryo datasets are overwhelmingly Caucasian. Collecting ethnically diverse datasets would improve global generalizability.

From Computer to Laboratory

The critical next step is in vitro validation: microinjecting compounds into cryopreserved embryos, monitoring development, sequencing them, and comparing results with in silico predictions to fine-tune the model.

Beyond Human Embryos

Adapting the Digital Embryo to other species could introduce multi-omic embryo analysis into conservation breeding, improve non-human IVF, and combat the extinction of endangered species.

Dr. Stewart Russell

In the summer of 2025, I reached out to Dr. Russell for mentorship. He held my computational methodology to the highest standards, taught me the foundations of reproductive biology, and guided this project from an idea into a framework I'm proud of. Beyond the science, he taught me to trust myself, to work hard, and above all, to enjoy the process. I’ll forever carry that with me. We're now preparing manuscripts for publication together.

The Digital Research Alliance of Canada

The Digital Research Alliance of Canada, via Dr. Russell’s sponsorship through Wilfrid Laurier University, provided access to the Nibi high-performance computing cluster. This made the computational work required to create The Digital Embryo possible.

My parents

My family has always supported my goals. They have dedicated their lives to my life, and I believe there is no nobler pursuit than that.

Thank you, everyone. Truly.

[1] Behi Shahreza, P. et al. (2026). Visible light, oxidative stress, and reproduction: The protective role of melatonin. Int. J. Gynecol. Obstet., 00:1-11. https://doi.org/10.1002/ijgo.70882

[2] Bergen, V. et al. (2020). Generalizing RNA velocity to transient cell states through dynamical modeling. Nat. Biotechnol., 38, 1408–1414. https://doi.org/10.1038/s41587-020-0591-3

[3] Bermejo-Álvarez, P. et al. (2010). Low oxygen tension during IVM improves bovine oocyte competence and enhances anaerobic glycolysis. Reprod. Biomed. Online., 20, 341-349. https://doi.org/10.1016/j.rbmo.2009.12.006

[4] Berntsen, J. et al. (2022). Robust and generalizable embryo selection based on artificial intelligence and time-lapse image sequences. PLoS One 17(2): e0262661. https://doi.org/10.1371/journal.pone.0262661

[5] Bhide, P. et al. (2024). Clinical effectiveness and safety of time-lapse imaging systems for embryo incubation and selection in in-vitro fertilisation treatment (TILT): a multicentre, three-parallel-group, double-blind, randomised controlled trial. Lancet, 404, 256–265. https://doi.org/10.1016/s0140-6736(24)00816-x

[6] Biondic, S. et al. (2024). Single-cell mRNA-sncRNA co-sequencing of preimplantation embryos. Methods Mol. Biol., 2767, 189-212. https://doi.org/10.1007/7651_2023_487

[7] Bloom, M. S. et al. (2011). Toxic trace metals and embryo quality indicators during in vitro fertilization (IVF). Reprod. Toxicol., 31, 164–170. https://doi.org/10.1016/j.reprotox.2010.11.011

[8] Brown, E. J. (2004). Analysis of cell cycle progression and genomic integrity in early lethal knockouts. Methods Mol. Biol., 280, 201-212. https://doi.org/10.1385/1-59259-788-2:201

[9] Canosa, S. et al. (2024). A novel machine-learning framework based on early embryo morphokinetics identifies a feature signature associated with blastocyst development. J. Ovarian Res., 17, 63. https://doi.org/10.1186/s13048-024-01376-6

[10] Carreira, J. & Zisserman, A. (2017). Quo Vadis, action recognition? A new model and the kinetics dataset. arXiv. https://doi.org/10.48550/arXiv.1705.07750

[11] Chatterjee, N. et al. (2023). Mixture and individual effects of benzene, toluene, and formaldehyde in zebrafish (Danio rerio) development: Metabolomics, epigenetics, and behavioral approaches. Environmental Toxicology and Pharmacology, 97, 104031. https://doi.org/10.1016/j.etap.2022.104031

[12] Chen, T. & Guestrin, C. (2016). XGBoost: A scalable tree boosting system. KDD, 785–794. https://doi.org/10.1145/2939672.2939785

[13] Chi, F. et al. (2020). Glycolysis-independent glucose metabolism distinguishes TE from ICM fate during mammalian embryogenesis. Dev. Cell, 53, 9–26.e4. https://doi.org/10.1016/j.devcel.2020.02.015

[14] Conde, P. et al. (2019). Effect of the hematopoietic growth factors erythropoietin and kit ligand on bovine oocyte in vitro maturation and developmental competence. Theriogenology, 123, 37–44. https://doi.org/10.1016/j.theriogenology.2018.09.014

[15] Coticchio, G. et al. (2023). The destinies of human embryos reaching blastocyst stage between Day 4 and Day 7 diverge as early as fertilization. Hum. Reprod., 38, 1690–1699. https://doi.org/10.1093/humrep/dead136

[16] Dang, Y. et al. (2023). Functional profiling of stage-specific proteome and translational transition across human pre-implantation embryo development at a single-cell resolution. Cell Discov., 9, 10. https://doi.org/10.1038/s41421-022-00491-2

[17] Dehghani-Mohammadabadi, M. et al. (2014). Melatonin modulates the expression of BCL-xl and improve the development of vitrified embryos obtained by IVF in mice. J. Assist. Reprod. Genet., 31, 453–461. https://doi.org/10.1007/s10815-014-0172-9

[18] Doerksen, T. & Trasler, J. M. (1996). Developmental exposure of male germ cells to 5-azacytidine results in abnormal preimplantation development in rats. Biol. Reprod., 55, 1155–1162. https://doi.org/10.1095/biolreprod55.5.1155

[19] Dou, J. et al. (2024). Single-nucleotide variant calling in single-cell sequencing data with Monopogen. Nat. Biotechnol., 42, 803–812. https://doi.org/10.1038/s41587-023-01873-x

[20] El-Mouatassim, S. et al. (2007). APEX/Ref-1 (apurinic/apyrimidic endonuclease DNA-repair gene) expression in human and ascidian (Ciona intestinalis) gametes and embryos. Mol. Hum. Reprod., 13, 549–556. https://doi.org/10.1093/molehr/gam038

[21] Fueyo, R. et al. (2025). A human-specific regulatory mechanism revealed in a preimplantation model. Nature., 647, 238-247. https://doi.org/10.1038/s41586-025-09571-1

[22] Gatimel, N. et al. (2020). Need for choosing the ideal pH value for IVF culture media. J. Assist. Reprod. Genet., 37, 1019–1028. https://doi.org/10.1007/s10815-020-01726-5

[23] Ghafarizade, A. A. et al. (2024). The clinical pregnancy and live birth following transfer of one arrested embryo: a case report. Clin. J. Obstet. Gynecol., 7(4), 112-114. https://doi.org/10.29328/journal.cjog.1001175

[24] Gu, Y. et al. (2023). Elevated intracellular pH of zygotes during mouse aging causes mitochondrial dysfunction associated with poor embryo development. Molecular and Cellular Endocrinology., 574, 111991. https://doi.org/10.1016/j.mce.2023.111991

[25] Hamilton, M. et al. (2022). Assessing spermatozoal small ribonucleic acids and their relationship to blastocyst development in idiopathic infertile males. Sci. Rep., 12, 20010. https://doi.org/10.1038/s41598-022-24568-w

[26] Hamilton, M. et al. (2024). A comprehensive analysis of spermatozoal RNA elements in idiopathic infertile males undergoing fertility treatment. Sci. Rep., 14, 10316. https://doi.org/10.1038/s41598-024-60586-6

[27] Hecker, D. et al. (2023). The adapted Activity-By-Contact model for enhancer–gene assignment and its application to single-cell data. Bioinformatics, 39, btad062. https://doi.org/10.1093/bioinformatics/btad062

[28] Hernandez Mora, J. R. et al. (2023). Single-cell multi-omic analysis profiles defective genome activation and epigenetic reprogramming associated with human pre-implantation embryo arrest. Cell Rep., 42, 112100. https://doi.org/10.1016/j.celrep.2023.112100

[29] Huang, W. et al. (2025). Practical probabilistic model-based reinforcement learning by integrating dropout uncertainty and trajectory sampling. IEEE Trans. Neural Netw. Learn. Syst., 36, 12812–12826. https://doi.org/10.1109/tnnls.2024.3474169

[30] Jeong, P.-S. et al. (2020). Chaetocin improves pig cloning efficiency by enhancing epigenetic reprogramming and autophagic activity. Int. J. Mol. Sci., 21(14), 4836. https://doi.org/10.3390/ijms21144836

[31] Kamijo, S. et al. (2022). MicroRNAs secreted by human preimplantation embryos and IVF outcome. Reprod. Biol. Endocrinol., 20, 130. https://doi.org/10.1186/s12958-022-00989-0

[32] Kamimoto, K. et al. (2023). Dissecting cell identity via network inference and in silico gene perturbation. Nature, 614, 742–751. https://doi.org/10.1038/s41586-022-05688-9

[33] Karmach, O. et al. (2022). Embryonic exposure to cigarette smoke extract impedes skeletal development and evokes craniofacial defects in zebrafish. Int. J. Mol. Sci., 23(17), 9904. https://doi.org/10.3390/ijms23179904

[34] Katušić-Bojanac, A. et al. (2022). Valproate Targets Mammalian Gastrulation Impairing Neural Tissue Differentiation and Development of the Placental Source In Vitro. Int. J. Mol. Sci., 23, 8861. https://doi.org/10.3390/ijms23168861

[35] Kuleshov, M. V. et al. (2016). Enrichr: a comprehensive gene set enrichment analysis web server 2016 update. Nucleic Acids Res., 44, W90–W97. https://doi.org/10.1093/nar/gkw377

[36] La Manno, G. et al. (2018). RNA velocity of single cells. Nature, 560, 494–498. https://doi.org/10.1038/s41586-018-0414-6

[37] Lange, M. et al. (2022). CellRank for directed single-cell fate mapping. Nat. Methods, 19, 159–170. https://doi.org/10.1038/s41592-021-01346-6

[38] Lawrenz, B. et al. (2024). Ongoing pregnancy rates in single euploid frozen embryo transfers remain unaffected by female age: a retrospective study. Reprod. Biomed. Online, 49, 104074. https://doi.org/10.1016/j.rbmo.2024.104074

[39] Leng, L. et al. (2019). Single-cell transcriptome analysis of uniparental embryos reveals parent-of-origin effects on human preimplantation development. Cell Stem Cell, 25, 697–712.e6. https://doi.org/10.1016/j.stem.2019.09.004

[40] Lesoine, B. & Regidor, P.-A. (2016). Prospective randomized study on the influence of myoinositol in PCOS women undergoing IVF in the improvement of oocyte quality, fertilization rate, and embryo quality. Int. J. Endocrinol., 2016, 4378507. https://doi.org/10.1155/2016/4378507

[41] Li, H. et al. (2025). Embryo-fetal developmental toxicity and toxicokinetics studies of YWS20045 orally administered to pregnant rats. Toxicology Letters., 413, 111749. https://doi.org/10.1016/j.toxlet.2025.111749

[42] Li, Z. et al. (2024). Exposure to volatile organic compounds induces cardiovascular toxicity that may involve DNA methylation. Toxicology., 501, 153705. https://doi.org/10.1016/j.tox.2023.153705

[43] Liu, L. et al. (2019). An integrated chromatin accessibility and transcriptome landscape of human pre-implantation embryos. Nat. Commun., 10, 364. https://doi.org/10.1038/s41467-018-08244-0

[44] Lopez, R. et al. (2018). Deep generative modeling for single-cell transcriptomics. Nat. Methods, 15, 1053–1058. https://doi.org/10.1038/s41592-018-0229-2

[45] Lundberg, S. & Lee, S.-I. (2017). A unified approach to interpreting model predictions. arXiv. https://doi.org/10.48550/arXiv.1705.07874

[46] Meistermann, D. et al. (2021). Integrated pseudotime analysis of human pre-implantation embryo single-cell transcriptomes reveals the dynamics of lineage specification. Cell Stem Cell, 28, 1625–1640.e6. https://doi.org/10.1016/j.stem.2021.04.027

[47] Milewski, R. et al. (2015). A predictive model for blastocyst formation based on morphokinetic parameters in time-lapse monitoring. J. Assist. Reprod. Genet., 32, 571–579. https://doi.org/10.1007/s10815-015-0440-3

[48] Nasr-Esfahani, F. et al. (2025). Differential effects of folic acid and 5-Methyltetrahydrofolate on mice oocytes and embryo development in vitro and in vivo: implications for reproductive health. The FASEB Journal., 39, e71090. https://doi.org/10.1096/fj.202501768R

[49] Nielsen, B. S. et al. (2025). Paracetamol (N-acetyl-para-aminophenol) disrupts early embryogenesis by cell cycle inhibition. Hum. Reprod., 40, 1860-1876. https://doi.org/10.1093/humrep/deaf116

[50] Orozco-Hernández, J. M. et al. (2022). Effects of oxidative stress induced by environmental relevant concentrations of fluoxetine on the embryonic development on Danio rerio. Science of The Total Environment., 807, 151048. https://doi.org/10.1016/j.scitotenv.2021.151048

[51] Otsuki, J. et al. (2007). Peroxidation of mineral oil used in droplet culture is detrimental to fertilization and embryo development. Fertil. Steril., 88, 741-743. https://doi.org/10.1016/j.fertnstert.2006.11.144

[52] O'Neill, C. (2005). The role of PAF in embryo physiology. Hum. Reprod. Update, 11, 215–228. https://doi.org/10.1093/humupd/dmi003

[53] Paloviita, P. et al. (2021). Small RNA expression and miRNA modification dynamics in human oocytes and early embryos. Genome Res., 31, 1474–1485. https://doi.org/10.1101/gr.268193.120

[54] Petersen, B. M. et al. (2016). Development of a generally applicable morphokinetic algorithm capable of predicting the implantation potential of embryos transferred on Day 3. Hum. Reprod., 31, 2231–2244. https://doi.org/10.1093/humrep/dew188

[55] Petropoulos, S. et al. (2016). Single-cell RNA-seq reveals lineage and X chromosome dynamics in human preimplantation embryos. Cell., 165(4), 1012–1026. https://doi.org/10.1016/j.cell.2016.03.023

[56] Pinto, S. et al. (2024). Metabolomics integration in assisted reproductive technologies for enhanced embryo selection beyond morphokinetic analysis. Int. J. Mol. Sci., 25, 491. https://doi.org/10.3390/ijms25010491

[57] Pontis, J. et al. (2022). Primate-specific transposable elements shape transcriptional networks during human development. Nat. Commun., 13, 7178. https://doi.org/10.1038/s41467-022-34800-w

[58] Qian, J. et al. (2020). Impacts of Caffeine during Pregnancy. Trends in Endocrinology & Metabolism., 31, 218-227. https://doi.org/10.1016/j.tem.2019.11.004

[59] Rai, A. et al. (2025). Dynamic proteome landscape during preimplantation human embryo development and trophectoderm stem cell-differentiation. Proteomics, 25, 72–89. https://doi.org/10.1002/pmic.70017

[60] Rajesh, V. & Divya, P. K. (2023). Embryonic exposure to decitabine induces multiple neural tube defects in developing zebrafish. Fish Physiol. Biochem., 49, 1357–1379. https://doi.org/10.1007/s10695-023-01261-x

[61] Rodriguez-Quiroz, R. & Valdebenito-Maturana, B. (2022). SoloTE for improved analysis of transposable elements in single-cell RNA-Seq data using locus-specific expression. Commun. Biol., 5, 1063. https://doi.org/10.1038/s42003-022-04020-5

[62] Rose, R. D. et al. (2020). The BlastGen study: a randomized controlled trial of blastocyst media supplemented with granulocyte-macrophage colony-stimulating factor. Reprod. Biomed. Online, 40, 645–652. https://doi.org/10.1016/j.rbmo.2020.01.011

[63] Rubner, Y. et al. (2000). The earth mover's distance as a metric for image retrieval. Int. J. Comput. Vis., 40, 99–121. https://doi.org/10.1023/a:1026543900054

[64] Russell, S. J. et al. (2020). Comprehensive profiling of Small RNAs in human embryo-conditioned culture media by improved sequencing and quantitative PCR methods. Syst. Biol. Reprod. Med., 66, 129–139. https://doi.org/10.1080/19396368.2020.1716108

[65] Russell, S. J. et al. (2024). An atlas of small non-coding RNAs in human preimplantation development. Nat. Commun., 15, 8634. https://doi.org/10.1038/s41467-024-52943-w

[66] Schiebinger, G. et al. (2019). Optimal-transport analysis of single-cell gene expression identifies developmental trajectories in reprogramming. Cell, 176, 928–943.e1-e22. https://doi.org/10.1016/j.cell.2019.02.026

[67] Sha, Q.-Q. et al. (2020). Dynamics and clinical relevance of maternal mRNA clearance during the oocyte-to-embryo transition in humans. Nat. Commun., 11, 4917. https://doi.org/10.1038/s41467-020-18680-6

[68] Sjoblom, C. et al. (1999). Granulocyte-macrophage colony-stimulating factor promotes human blastocyst development in vitro. Hum. Reprod., 14, 3069–3076. https://doi.org/10.1093/humrep/14.12.3069

[69] Slattery, M. E. et al. (2025). Methotrexate use in the management of ectopic pregnancies: a retrospective study. Cureus., 17, e95826. https://doi.org/10.7759/cureus.95826

[70] Song, Y. et al. (2021). Solving inverse problems in medical imaging with score-based generative models. arXiv. https://doi.org/10.48550/arXiv.2111.08005

[71] Strozzi, F. et al. (2019). Scalable workflows and reproducible data analysis for genomics. Methods Mol. Biol., 1910, 723–745. https://doi.org/10.1007/978-1-4939-9074-0_24

[72] Subramanian, A. et al. (2005). Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl Acad. Sci. USA, 102, 15545–15550. https://doi.org/10.1073/pnas.0506580102

[73] Sukhn, C. et al. (2021). Paternal exposure to non-essential heavy metal affects embryo cleavage and implantation in intracytoplasmic sperm injection (ICSI) cycles: evidence for a paradoxical effect. Reprod. Sci., 28, 2550–2561. https://doi.org/10.1007/s43032-021-00510-9

[74] Thompson, J. G. et al. (2000). Effect of inhibitors and uncouplers of oxidative phosphorylation during compaction and blastulation of bovine embryos cultured in vitro. J. Reprod. Fertil., 118, 47–55. https://doi.org/10.1530/reprod/118.1.47

[75] Tong, A. et al. (2023). Simulation-free Schrödinger bridges via score and flow matching. arXiv. https://doi.org/10.48550/arXiv.2307.03672

[76] Trincado, J. L. et al. (2018). SUPPA2: fast, accurate, and uncertainty-aware differential splicing analysis across multiple conditions. Genome Biol., 19, 40. https://doi.org/10.1186/s13059-018-1417-1

[77] Tulay, P. et al. (2017). Differential expression of parental alleles of BRCA1 in human preimplantation embryos. Eur. J. Hum. Genet., 25, 37–42. https://doi.org/10.1038/ejhg.2016.121

[78] Vitale, S. G. et al. (2016). How to achieve high-quality oocytes? The key role of myo-inositol and melatonin. Int. J. Endocrinol., 2016, 4987436. https://doi.org/10.1155/2016/4987436

[79] Wang, G. et al. (2025). scMODAL: a general deep learning framework for comprehensive single-cell multi-omics data alignment with feature links. Nat. Commun., 16, 4994. https://doi.org/10.1038/s41467-025-60333-z

[80] Wang, S. et al. (2024). A single-cell transcriptome atlas of human euploid and aneuploid blastocysts. Nat. Genet., 56, 1468–1481. https://doi.org/10.1038/s41588-024-01788-6

[81] Wang, T. et al. (2024). Single-cell multi-omics profiling of human preimplantation embryos identifies cytoskeletal defects during embryonic arrest. Nat. Cell Biol., 26, 263–277. https://doi.org/10.1038/s41556-023-01328-0

[82] Wyck, S. et al. (2018). Oxidative stress in sperm affects the epigenetic reprogramming in early embryonic development. Epigenetics Chromatin, 11, 60. https://doi.org/10.1186/s13072-018-0224-y

[83] Xu, S. et al. (2025). Association of combined exposure to air pollutants during ovarian stimulation, follicular fluid metabolites, and IVF/ICSI outcomes: A prospective cohort study. Journal of Hazardous Materials., 499, 140067. https://doi.org/10.1016/j.jhazmat.2025.140067

[84] Yan, L. et al. (2013). Single-cell RNA-Seq profiling of human preimplantation embryos and embryonic stem cells. Nat. Struct. Mol. Biol., 20, 1131–1139. https://doi.org/10.1038/nsmb.2660

[85] Yang, Y. et al. (2022). Metabolic and epigenetic dysfunctions underlie the arrest of in vitro fertilized human embryos in a senescent-like state. PLoS Biol., 20, e3001682. https://doi.org/10.1371/journal.pbio.3001682

[86] Zhang, H. et al. (2023). Stable maternal proteins underlie distinct transcriptome, translatome, and proteome reprogramming during mouse oocyte-to-embryo transition. Genome Biol., 24, 166. https://doi.org/10.1186/s13059-023-02997-8

[87] Zhang, H. et al. (2025). GMNN mutations cause female infertility characterized by preimplantation embryo arrest through regulating DNA re-replication. Sci. China Life Sci., 68, 2420–2430. https://doi.org/10.1007/s11427-024-2837-9

[88] Zhang, S. et al. (2015). Effect of DNMT inhibitor on bovine parthenogenetic embryo development. Biochemical and Biophysical Research Communications, 466, 505–511. https://doi.org/10.1016/j.bbrc.2015.09.060

[89] Zhang, Y.-M. et al. (2018). Effects of histone methyltransferase inhibitor chaetocin on histone H3K9 methylation of cultured ovine somatic cells and development of preimplantation cloned embryos. Reprod. Toxicol., 79, 124–131. https://doi.org/10.1016/j.reprotox.2018.06.006

[90] Zhang, Y.-W. et al. (2026). A homozygous variant in HFM1 causes preimplantation embryo developmental arrest by disrupting zygotic genome activation. Hum. Reprod., 41(2), 296-309. https://doi.org/10.1093/humrep/deaf238

[91] Zhu, W. et al. (2025). Comparative proteomic landscapes elucidate human preimplantation development and failure. Cell, 188, 814–831.e21. https://doi.org/10.1016/j.cell.2024.12.028

[92] Ziebe, S. et al. (2013). A randomized clinical trial to evaluate the effect of granulocyte-macrophage colony-stimulating factor (GM-CSF) in embryo culture medium for in vitro fertilization. Fertil. Steril., 99, 1600–1609. https://doi.org/10.1016/j.fertnstert.2012.12.043

[93] Zuev, D. et al. (2025). IVF and thermal manipulation at the first cleavage stage alter offspring circadian phenotype, sleep, and brain epigenetics. Int. J. Mol. Sci., 26, 10360. https://doi.org/10.3390/ijms262110360