Biology subjects

Galvin, J.

Publications and source records attributed to Galvin, J..

4 recordsLinked to original sources

bioRxiv2026

Synthetic MPRA perturbation supervision improves variant-effect prediction in sequence-to-function models

Deep learning sequence-to-function (S2F) models learn to predict functional readouts, such as gene expression, from genetic sequence. However, these models are trained predominantly on the reference genome and struggle to accurately predict the effect of genetic variation on gene expression. Massively parallel reporter assays (MPRAs) explore the direct impact of genetic variation on gene expression, but saturation-mutagenesis MPRAs that measure variant effects directly remain limited, thereby restricting their use for fine-tuning S2F models. In this work, we investigate fine-tuning Borzoi, a state-of-the-art S2F model, on model-generated synthetic MPRA variant-effect data as a scalable alternative to overcome limits on the availability of supervised training data for genetic variation. First, we trained an ensemble of 16 models on 110,359 K562 MPRA measurements. Then, we distilled their predictions into a single student model using MPRA elements associated with 1,000 genes. The distilled model was subsequently used to generate synthetic variant-effect labels across increasing numbers of genes (50 to 1,000), with 600 variants per gene, to fine-tune Borzoi. Performance improved as we included more genes in training. In data-limited settings, training on more genes with fewer variants per gene performed better than training on fewer genes with more variants per gene, given the same number of total variants. Compared to pre-trained Borzoi, Borzoi fine-tuned on 600,000 synthetic samples across 1,000 genes increased Pearson correlation from 0.28 to 0.54 on CAGI5 promoter variants and from 0.39 to 0.46 on 903 GTEx fine-mapped whole-blood expression quantitative trait loci (eQTLs). Importantly, these improvements were achieved while preserving the pre-trained performance across all 7,611 output tracks. Furthermore, fine-tuned Borzoi recovered regulatory motifs identified by the MPRA model in genes unseen during training, suggesting that learned regulatory features can generalize across loci. Together, these results provide a proof of concept that model-generated synthetic supervision can improve variant effect prediction in S2F models while preserving their original pre-trained capabilities, offering a scalable strategy toward capturing the impact of genetic variation on gene regulation.

genomics↗

bioRxiv2025

Fine-tuning sequence to function deep learning models on large-scale proteomic data improves the accuracy of variant effect prediction

Fine-tuning sequence to function models has shown promise for variant effect prediction, but accuracy and generalization to unseen genes and unseen individuals remains a standing challenge. We fine-tuned Borzoi on 54,219 individuals and 2,923 circulating plasma proteins from the UK Biobank Plasma Proteomic Project. Across 150 single-gene models where the genes had a range of cis-heritability we observed that the fine-tuned Borzoi model improved variant effect prediction for 86% of the genes compared to an Elastic Net baseline model. We demonstrated that the improved prediction stems from increased sample size which provides tremendous amounts of rare genetic variants (MAF < 0.01) to the training data. Masking rare and uncommon variants nullified improved performance of fine-tuned Borzoi and we showed that fine-tuned Borzoi highly weights rare variants (MAF < 0.01) while the Elastic Net model highly weights common variants (MAF > 0.05) that are enriched for regulatory regions. We evaluated the generalizability of our model on a fine-tuned Borzoi model trained jointly on varying numbers of genes and observed that these models consistently outperform the pre-trained Borzoi model, the single-gene models yield more accurate results. Together this work demonstrates the importance of including larger sample sizes and rare variants in sequence to function models for variant effect prediction and demonstrates feasibility that these models are capable of highly accurate variant effect prediction.

genetics↗

bioRxiv2021

Genetic coordination of sperm morphology and seminal fluid proteins promotes male reproductive success in Drosophila melanogaster

Spermatozoal morphology is highly variable both among and within species and in ways that can significantly impact fertilization success. In Drosophila melanogaster, paternity success depends on sperm length of both competing males and length of the females primary sperm storage organ. We found that genes upregulated in long sperm testes are enriched for lncRNAs and seminal fluid proteins (Sfps). Transferred in seminal fluid to the female during mating, Sfps are secreted by the male accessory glands (AG) and affect female remating rate, physiology, and behavior with concomitant advantages for male reproductive success. Despite being upregulated in long sperm testes, they have no known function in testis tissue. We found that Sex Peptide and ovulin (Acp26Aa) knockouts resulted in shorter sperm, suggesting that Sfps may regulate sperm length during spermatogenesis. However, knockout of AG function did not affect sperm length, suggesting that AG expression has no influence on spermatogenic processes. We also found that long sperm males are better able to delay female remating, suggesting higher Sfp expression in AG. These results might suggest that long sperm males have a double advantage in sperm competition by both delaying female remating, likely through transfer of more Sfps, and by resisting sperm displacement. However, we also found that this extra advantage does not necessarily translate to more progeny or higher paternity success. Thus, we found that multiple components of the ejaculate coordinate to promote male reproductive success at different stages of reproduction, but the realized fitness advantages in sperm competition are uncertain. Significance StatementThe ejaculate is comprised of sperm produced in the testis and seminal fluid primarily produced in the male accessory glands (AG). These complementary components are both critical for male reproductive success, but they are largely considered to be functionally, genetically, and developmentally independent. In a quest to understand genetic mechanisms of sperm length variation, we found that genes upregulated in long sperm testes are enriched for lncRNAs and seminal fluid proteins (Sfps). Knockout of two Sfps, Sex Peptide and ovulin, results in shorter sperm, though knockout of AG function has no effect. Moreover, long sperm males delay female remating longer. These results suggest sophisticated testis-AG coordination that amplifies male reproductive success, with implications for evolutionary integration of sexually selected traits.

genetics↗

bioRxiv2020

Assortative mating, autistic traits, empathizing, and systemizing

It has been suggested that the children of parents with particular interests and aptitude for understanding systems via input-operation-output rules (i.e. systemizing) are at increased likelihood of developing autism. Furthermore, assortative mating (i.e. a non-random pattern in which individuals are more likely to pair with others who are similar to themselves) is hypothesised to occur in relation to systemizing, and so romantic couples may be more similar on this variable than chance would dictate. However, no published study has yet tested this hypothesis. We therefore examined intra-couple correlations for a measure of autistic traits (Autism Spectrum Quotient [AQ]), self-report measures of empathizing (Empathy Quotient [EQ]), and systemizing (Systemizing Quotient-Revised [SQ-R]), as well as the Reading the Mind in the Eyes Test (RMET) and Embedded Figures Task (EFT). We observed positive intra-couple correlations of small-to-medium magnitude for all measures except EQ. Further analyses suggest that these effects are attributable to people pairing with those who are more similar to themselves than chance (initial assortment) rather than becoming more alike over the course of a relationship (convergence), and to seeking out self-resembling partners (active assortment) rather than pairing in this manner due to social stratification increasing the likelihood of similar people meeting in the first place (social homogamy). Additionally, we found that the difference in scores for the AQ, SQ-R, RMET and EFT of actual couples were smaller (i.e. more similar) than the average difference scores calculated from all other possible male-female pairings within the dataset. The current findings therefore provide clear evidence in support of the assortative mating theory of autism.

evolutionary biology↗