Adam C. Siepel

The paper identifies 17 human thymic epithelial cell subsets using single-cell data and shows they share a common ancestor and regulatory logic, with some conservation in mice.

The paper introduces scPFA, a lineage-informed factor analysis method, to uncover heritable gene expression programs from single-cell lineage-tracing data.

The paper introduces a model-based framework to compare promoter-proximal pausing dynamics across biological contexts, revealing kinetic and distributional variations in pausing.

This paper introduces a variational inference method for Bayesian phylogenetic reconstruction, showing faster speeds than traditional Bayesian methods on real-world datasets.

LaVOUS models gene-expression dynamics along cell lineages, identifying lineage-associated changes in cancer, B cells, and brain development.

This study uses a domain-adaptive neural network to detect selective sweeps in ancient Eurasian genomes, finding that hard sweeps at loci related to neuronal, reproductive, pigmentation, and signaling traits persisted despite admixture.

This paper introduces BEAM, a Bayesian framework for inferring tissue migration histories in metastatic cancer from lineage tracing data, showing improved performance over existing methods.

GHIST 2024 evaluates population genetic inference methods through four demographic history challenges, revealing the dominance of site frequency spectrum-based methods and the potential of flexible models for complex histories.

The paper examines genome evolution in maize and related grasses, revealing variable outcomes of polyploidy across species.

The paper introduces ASSESS, a method to estimate the distribution of fitness effects from GWAS summary statistics, demonstrating its accuracy through simulations and application to human trait data.

This study maps the transcriptional responses of murine and human mammary organoids to female hormones, revealing molecular dynamics across species.

The paper investigates DNA-sequence and epigenomic factors influencing local transcription elongation rates.

The paper applies domain adaptation to supervised machine learning models trained on simulated population genetic data to improve performance on real-world data.

This study evaluates AlphaMissense's ability to classify ASNS variants in ASNSD, finding it accurately identifies most known pathogenic variants and provides a useful tool for variant classification.

The paper develops a probabilistic model to estimate transcription initiation and pause-escape rates from nascent RNA sequencing data, showing pause-escape is often more rate-limiting than previously thought.

This study uses long-read sequencing to analyze immune and cancer-related gene evolution in bats, finding gene contractions, expansions, and positive selection linked to their unique traits.

The paper shows that RNA Pol II pausing is essential for proper pluripotency transitions in mouse embryos by maintaining balance between gene expression states.

ACE introduces a probabilistic model to assess gene essentiality in CRISPR screens, identifying differential essentiality across samples and uncovering genotype-specific essentiality.

SIA detects and quantifies positive selection using deep learning on the ancestral recombination graph, showing performance comparable to existing methods and identifying novel selection signals in human and bird populations.

The paper develops a multiple genome alignment workflow for plants, showing that k-mer masking and parameter tuning improve alignment of functional regions.

The paper introduces a method to estimate RNA half-lives using PRO-seq and RNA-seq data, finding correlations between RNA stability and splicing, miRNA binding, DNA methylation, and histone modifications.

This paper introduces a probabilistic model for analyzing nascent RNA sequencing data to estimate transcription kinetics and perform likelihood ratio tests across genes and conditions.

The paper introduces LASSIE, a model to estimate allele-specific fitness effects across human protein-coding sequences, showing its relevance to evolution and disease.

The paper quantifies how much information human epigenomic data provides about genomic function using evolutionary models and finds that most genetic variation reflects neutral processes, with only a small fraction constrained by selection.

The paper compares nascent transcription in primates to identify evolutionary patterns in coding and non-coding transcription, finding that transcription factor binding sites and adaptive substitutions drive species-specific differences.

Scikit-ribo provides accurate A-site prediction and translation efficiency estimation from Riboseq and RNAseq data, correcting biases and improving TE estimation for over 2000 genes in S. cerevisiae.

The paper generates a comprehensive sRNA catalog across 12 Drosophila species to study miRNA diversity, evolution, and functional variation.

This study uses PRO-seq to analyze the rapid transcriptional response to celastrol in K562 cells, revealing two waves of transcriptional changes and identifying key transcription factors involved.

This paper introduces LINSIGHT, a method that improves prediction of deleterious noncoding variants by combining neural networks and molecular evolution models, outperforming existing methods in identifying disease-associated variants.