Georg Seelig

The paper introduces CytED, a database that transforms cytokine literature into a resource for experimental analysis, identifying unexpected biological effects and guiding perturbation experiments.

The paper introduces mapSPLiT, a pooled CRISPRa/i screening platform for bacteria, linking perturbations to single-cell transcriptomes and expanding regulatory networks in E. coli and P. putida.

The paper creates a Human Cytokine Dictionary using single-cell transcriptomes of PBMCs stimulated with cytokines, revealing cell-type-specific responses and communication networks.

The paper develops COMPASS, a method to assess splicing variant effects across cell types, identifying disease-associated variants and cell type-specific splicing programs.

The paper designs cell type-specific enhancers using deep learning models trained on chromatin accessibility data and validates their function across human and mouse cell types.

GAME provides a standardized framework for benchmarking genomics models across tasks, enabling consistent evaluation and community contributions.

This study uses iterative deep learning to design human enhancers with cell type-specific activity, validated through experimental testing and perturbation experiments.

The paper designs overlapping genes using deep generative models of protein sequences and shows high success rates in creating functional proteins in alternate reading frames.

The paper identifies a nuclear RNA degradation code (NRDC) that recognizes transcripts with both a 5' splice site and a poly(A) junction, leading to their degradation and contributing to disease.

This study uses MPRA and polysome profiling to analyze 5' UTR regulatory features in zebrafish embryos, identifying motifs and factors influencing ribosome recruitment and translation dynamics.

The paper profiles single-cell transcriptomes of Bacteroides fragilis infected with a phage, revealing combinatorial phase-variable expression of defense systems that enable super-resistant cells to avoid phage infection without mutations.

The paper develops a model to predict mRNA stability based on microRNA expression data and applies it to design microRNA-responsive constructs with specific activity patterns across cell types.

This study creates a single-cell multi-omics atlas of zebrafish embryogenesis and identifies transcription factors that drive cell-type-specific chromatin landscapes and instant differentiation.

The paper uses deep learning to design optimized 5'UTRs for mRNA-delivered gene editing, showing improved editing efficiency across cell types and targets.

The paper identifies sequence-specific sensitivity to JTE-607 in mRNA 3' processing by analyzing over 260,000 sequence variants and developing a predictive model.

This paper develops thermoresponsive microcapsules for repeated random access of DNA data, demonstrating improved performance over non-compartmentalized storage.

The paper introduces MP3-seq, a scalable method for measuring protein-protein interactions, and shows it can quantify over 100,000 interactions.

The paper introduces APARENT2, a neural network model for predicting polyadenylation effects, and finds gain-of-function mutations linked to Autism Spectrum Disorder.

The paper designs modular yeast consortia for synthetic biology, showing that consortia dynamics can be predicted from individual strain models and optimized to achieve target behaviors.

This paper introduces Scrambler networks, a method for interpreting sequence-based neural networks by learning stochastic masks to identify salient sequence positions and explain model predictions.

CellMeSH automatically annotates cell types from scRNA-seq data using a probabilistic method and a literature-based database of gene-cell type associations.

This study maps the spatial and single-cell transcriptional landscape of human cerebellar development, revealing cell-type-specific gene expression and implications for neurological disorders.

This paper quantifies molecular bias in DNA data storage by analyzing synthesis and PCR processes, identifying their contributions to copy number unevenness and proposing a statistical model to optimize storage efficiency.

UNCURL-App provides an interactive tool for scRNA-seq analysis, integrating cell type databases and allowing user-driven re-analysis of clusters and pathways.

This paper demonstrates a DNA-based reaction-diffusion system that generates programmable, tunable patterns at the centimeter scale through synthetic DNA complexes embedded in hydrogel.

This paper introduces Deep Exploration Networks (DENs) to generate high-fitness DNA sequences, demonstrating improved performance in engineering RNA isoforms compared to gradient ascent methods.

The paper develops a model to predict the effect of 5' UTR variants on translation using experimental data and deep learning, showing its application to disease-associated SNVs.

This paper introduces APARENT, a deep learning model that predicts alternative polyadenylation from DNA sequence, outperforming models trained on endogenous data and identifying genetic variants impacting APA in disease contexts.

This study investigates how microRNA-mediated incoherent feedforward loops suppress extrinsic noise, showing that translational inhibition expands noise rejection regimes and can buffer or amplify gene product fluctuations depending on time scales.

This paper introduces PIPER, a method to identify gene network perturbations in single-cell RNA-seq data, showing it outperforms existing methods on simulated data and identifies known regulators in real datasets.