Biology subjects

Tran, L. N.

Publications and source records attributed to Tran, L. N..

7 recordsLinked to original sources

bioRxiv2025

ConfuseNN: Interpreting convolutional neural network inferences in population genomics with data shuffling

Convolutional neural networks (CNNs) have become powerful tools for population genomic inference, yet understanding which genomic features drive their performance remains challenging. We introduce ConfuseNN, a method that systematically shuffles input haplotype matrices to disrupt specific population genetic features and evaluate their contribution to CNN performance. By sequentially removing signals from linkage disequilibrium, allele frequency, and other population genetic patterns in test data, we evaluate how each feature contributes to CNN performance. We applied ConfuseNN to three published CNNs for demographic history and selection inference, confirming the importance of specific data features and identifying limitations of network architecture and of simulated training and testing data design. ConfuseNN provides an accessible biologically motivated framework for interpreting CNN behavior across different tasks in population genetics, helping bridge the gap between powerful deep learning approaches and traditional population genetic theory.

genetics↗

bioRxiv2025

Accessible, realistic genome simulation with selection using stdpopsim

Selection is a fundamental evolutionary force that shapes patterns of genetic variation across species. However, simulations incorporating realistic selection along heterogeneous genomes in complex demographic histories are challenging, limiting our ability to benchmark statistical methods aimed at detecting selection and to explore theoretical predictions. stdpopsim is a community-maintained simulation library that already provides an extensive catalog of species-specific population genetic models. Here we present a major extension to the stdpopsim framework that enables simulation of various modes of selection, including background selection, selective sweeps, and arbitrary distributions of fitness effects (DFE) acting on annotated subsets of the genome (for instance, exons). This extension maintains stdpopsims core principles of reproducibility and accessibility while adding support for species-specific genomic annotations and published DFE estimates. We demonstrate the utility of this framework by comparing methods for demographic inference, DFE estimation, and selective sweep detection across several species and scenarios. Our results demonstrate the robustness of demographic inference methods to selection on linked sites, reveal the sensitivity of DFE-inference methods to model assumptions, and show how genomic features, like recombination rate and functional sequence density, influence power to detect selective sweeps. This extension to stdpopsim provides a powerful new resource for the population genetics community to explore the interplay between selection and other evolutionary forces in a reproducible, user-friendly framework.

evolutionary biology↗

bioRxiv2025

Epigenetic priming of neural progenitors by Notch enhances Sonic hedgehog signaling and establishes gliogenic competence

The remarkable cell diversity of multicellular organisms relies on the ability of multipotent progenitor cells to generate distinct cell types at the right times and locations during embryogenesis. A key question is how progenitors establish competence to respond to the different environmental signals required to produce specific cell types at critical developmental timepoints. We addressed this in the mouse developing forebrain, where neural progenitor cells must switch from producing neurons to making oligodendrocytes in response to increased Sonic Hedgehog (SHH) signaling during late embryogenesis. We show that progenitor responses to SHH are regulated by Notch signaling, thus permitting proper timing of the neuron-oligodendrocyte switch. Notch activity epigenetically primes genes associated with the oligodendrocyte lineage and SHH pathway, enabling amplified transcriptional responses to endogenous SHH and robust oligodendrogenesis. These results reveal a critical role for Notch in facilitating progenitor competence states and influencing cell fate transitions at the epigenetic level.

developmental biology↗

bioRxiv2024

Modeling biases from low-pass genome sequencing to enable accurate population genetic inferences

Low-pass genome sequencing is cost-effective and enables analysis of large cohorts. However, it introduces biases by reducing heterozygous genotypes and low-frequency alleles, impacting subsequent analyses such as demographic history inference. We developed a probabilistic model of low-pass biases from the Genome Analysis Toolkit (GATK) multi-sample calling pipeline, and we implemented it in the population genomic inference software dadi. We evaluated the model using simulated low-pass datasets and found that it alleviated low-pass biases in inferred demographic parameters. We further validated the model by downsampling 1000 Genomes Project data, demonstrating its effectiveness on real data. Our model is widely applicable and substantially improves model-based inferences from low-pass population genomic data.

evolutionary biology↗

bioRxiv2024

Meningeal-derived retinoic acid regulates neurogenesis via suppression of Notch and Sox2

The meninges act as a regulator of brain development by secreting ligands that act on neural cells to regulate neurogenesis and neuronal migration. Meningeal-derived retinoic acid (RA) promotes neocortical neural progenitor cell cycle exit; however, the underlying molecular mechanism is unknown. Here, we used spatial transcriptomics and profiling of retinoic-acid receptor- (RAR) DNA binding in Foxc1-mutant embryos that lack meninges-derived ligands to identify the neurogenic transcriptional mechanisms of RA signaling in telencephalic neural progenitors. We determined that meningeal-derived RA controls neurogenesis by suppressing progenitor self-renewal pathways Notch signaling and the transcription factor Sox2. We show that RAR binds in the Sox2ot promoter, a long non-coding RNA that regulates Sox2 expression, and RA promotes Sox2ot expression in neocortical progenitors. Our findings elucidate a previously unknown mechanism of how meningeal RA coordinates neocortical development and insight into how defects in meningeal development may cause neurodevelopmental disorders.

neuroscience↗

bioRxiv2023

Computationally efficient demographic history inference from allele frequencies with supervised machine learning

Inferring past demographic history of natural populations from genomic data is of central concern in many studies across research fields. Previously, our group had developed dadi, a widely used demographic history inference method based on the allele frequency spectrum (AFS) and maximum composite likelihood optimization. However, dadis optimization procedure can be computationally expensive. Here, we developed donni (demography optimization via neural network inference), a new inference method based on dadi that is more efficient while maintaining comparable inference accuracy. For each dadi-supported demographic model, donni simulates the expected AFS for a range of model parameters then trains a set of Mean Variance Estimation neural networks using the simulated AFS. Trained networks can then be used to instantaneously infer the model parameters from future input data AFS. We demonstrated that for many demographic models, donni can infer some parameters, such as population size changes, very well and other parameters, such as migration rates and times of demographic events, fairly well. Importantly, donni provides both parameter and confidence interval estimates from input AFS with accuracy comparable to parameters inferred by dadis likelihood optimization while bypassing its long and computationally intensive evaluation process. donnis performance demonstrates that supervised machine learning algorithms may be a promising avenue for developing more sustainable and computationally efficient demographic history inference methods.

genomics↗

bioRxiv2020

Unraveling the transcriptional networks that drive oligodendrocyte fate specification in Sonic hedgehog-responsive neocortical progenitors

In the developing nervous system, progenitors first generate neurons before making astrocytes and oligodendrocytes. We previously showed that increased Sonic hedgehog (Shh) signaling in dorsal forebrain progenitors is important for their production of oligodendrocytes as neurogenesis winds down. Here, we analyzed single-cell RNA sequencing datasets to better understand how Shh controls this neuron-to-oligodendrocyte switch in the neocortex. We first identified Shh-responding progenitors using a dataset in which Shh was overexpressed in the mouse dorsal forebrain. Pseudotime trajectory inferences revealed a subpopulation committed to the oligodendrocyte precursor cell (OPC) lineage. Genes upregulated along this lineage defined a pre-OPC state, as cells transitioned from progenitors to OPCs. Using several datasets from wild-type mouse and human embryos at different ages, we confirmed a pre-OPC state preceding OPC emergence during normal development. Finally, we show that pre-OPCs are enriched for a gene regulatory network involving the transcription factor Ascl1. Genetic lineage-tracing demonstrated Ascl1+ dorsal progenitors primarily make oligodendrocytes. We propose a model in which Shh shifts the balance between opposing transcriptional networks toward an Ascl1 lineage, thereby facilitating the switch between neurogenesis and oligodendrogenesis.

developmental biology↗