David S. Fischer

The paper develops a framework combining perturbation screens and transfer learning to reconstruct signaling histories in cells and optimize differentiation protocols.

This study uses spatial transcriptomics to show that aging disrupts spatial and temporal coordination of folliculogenesis in mouse ovaries, leading to tissue disorganization and reproductive decline.

This paper introduces kai, an agentic AI that generates analyses in computational notebooks to align human and agentic reasoning in single-cell biology, showing improved robustness and reasoning capabilities.

This study benchmarks self-supervised learning techniques in single-cell genomics, finding masked autoencoders outperform contrastive methods in cell type prediction and data integration.

This study shows that using fragment counts instead of binarized data improves scATAC-seq analysis methods like clustering and batch integration.

This paper introduces scTab, a deep learning model for cross-tissue cell type annotation in single-cell RNA-seq, showing improved performance over linear models and demonstrating scalability with larger datasets.

This paper evaluates graph neural networks' ability to predict tissue-level phenotypes from spatial molecular data, finding that some phenotypes can be captured without spatial information and that overfitting limits representation learning.

The paper develops a high-sensitivity mass spectrometry workflow to quantify single-cell proteome changes upon perturbation, revealing a stable core proteome despite transcriptomic variability.

Squidpy extends Scanpy/AnnData with spatial neighbor graphs and a lazy-loading image container, enabling spatial statistics, ligand-receptor tests, image features, and ~10-fold faster neighborhood enrichment than Giotto.

epiScanpy enables analysis of single-cell DNA methylation and ATAC-seq data using scanpy workflows, showing epigenetic markers enrich cell type labels.

Sfaira provides a data and model zoo for single-cell RNA-seq, enabling fast reuse of pre-trained models and datasets across human and mouse tissues.

The paper introduces NCEM, a graph neural network method for modeling cell communication in spatial molecular profiling data, identifying niche-dependent gene expression and cell state variation across multiple tissues.

The paper develops deep learning models to predict T-cell antigen specificity from TCR CDR3 sequences and finds categorical antigen modeling outperforms joint sequence modeling.

Across 1,176,683 airway and lung cells from 22 datasets, ACE2 rises with age, in males and in smokers' multiciliated cells, and is lowest in young children.

This paper introduces methods to identify globally and locally relevant genes in nonlinear low-dimensional embeddings of single-cell RNA-seq data, demonstrating their versatility across datasets and embedding techniques.

The paper investigates how glucocorticoid receptor activation affects neuronal development in fetal human brain organoids using single-cell transcriptomics, finding cell-type specific impacts on differentiation and behavior-related gene expression.

MPRAnalyze provides a statistical framework for analyzing MPRA data to estimate regulatory sequence effects and compare differential activity across conditions.

The paper introduces pseudodynamics, a framework to model real-time cellular development using time-series and genetic knock-out data, applied to thymic T-cell maturation.

ImpulseDE2 improves differential expression analysis for time course sequencing data by modeling gene expression trajectories with an impulse model, outperforming existing methods in distinguishing transient and monotonic changes.