Ge Gao

This study evaluates 11 genomic language models and finds they offer limited advantages over a random baseline in regulatory decoding tasks.

CASCADE is a framework for causal regulatory inference and rational transcriptomic reprogramming design, enabling causal discovery and intervention design for targeted cells.

The paper introduces a metric to quantify gene expression predictability from histology images and identifies genes that can or cannot be predicted.

PASSAGE identifies phenotype-associated spatial signatures in spatial transcriptomics data using a deep learning framework.

DECIPHER generates high-fidelity disentangled embeddings for large-scale heterogeneous spatial omics data, outperforming existing methods and enabling global analysis.

This paper introduces SVEN, a hybrid sequence-oriented model that predicts tissue-specific transcriptomic impacts of structural variants across 200+ tissues and cell lines.

SLAT aligns heterogeneous spatial omics data by integrating omics profiles and spatial context, improving cell-typing resolution and regulatory inference.

The paper introduces Markonv, a convolutional layer that models inter-positional correlations using Markov transition matrices, showing effectiveness in analyzing omics data and decoding electrical signals.

This paper introduces GLUE, a framework for integrating unpaired single-cell multi-omics data using regulatory knowledge, showing improved accuracy and scalability across various tasks.

The paper introduces RiboCalc, a model that quantitatively predicts transcript coding ability in human cells, revealing context-dependent coding transcripts.

REVA is a curated database of over 11.8 million noncoding variants with expression-modulating potential, providing functional annotations and benchmarking computational tools for variant analysis.

The paper introduces a variable-convolutional layer for motif identification in omics data, showing improved performance over standard convolutional layers.

This paper introduces COSLIR, a method to directly reconstruct gene regulatory networks from time-stamped single-cell data without pseudo-time inference, showing good performance on real datasets.

Cell BLAST provides a tool for querying and annotating scRNA-seq data using a reference database and web interface.

The paper introduces GeACT, a genomic database for human mid-gestation fetus cells, linking gene expression to chromatin states and identifying key transcription factors.

The paper identifies over two million functional transcription factor interactions in 63 green plants using evolutionary footprints and a new algorithm.

This paper introduces expectation pooling, an interpretable CNN pooling method that improves DNA-protein binding prediction by combining probabilistic ideas with deep learning.

This paper introduces an exact transformation to convert CNN kernels to PWMs, enabling accurate motif identification and preserving CNN performance.