Constantin Ahlmann-Eltze

The paper benchmarks deep learning models against linear baselines for predicting gene perturbation effects and finds they do not outperform simple models.

LEMUR identifies gene expression changes across conditions using a continuous latent space, improving analysis of multi-condition single-cell RNA-seq data.

The paper compares transformation methods for single-cell RNA-seq data and finds that a simple log transformation with pseudo-count and PCA performs as well as more complex approaches.

The paper introduces a new R package for efficiently fitting Gamma-Poisson models to single-cell RNA-seq data, improving speed and accuracy over existing methods.

proDA models intensity-dependent missing values probabilistically within variance-moderated linear models to test differential protein abundance in label-free mass spectrometry without the false positives of ad hoc imputation.