Xinming Tu

Flexible neuronal participation within a reliable motor sequence

The paper shows that reliable motor sequences can maintain timing stability while allowing flexible neuronal participation in zebra finch songs.

A scalable approach to investigating sequence-to-function predictions from personal genomes

This paper introduces SAGE-net, a framework for training sequence-to-function models using personal genomes, showing improved gene expression prediction but limited generalization, while DNA methylation shows better generalization.

A modality gap in personal-genome prediction by sequence-to-function models

The paper evaluates AlphaGenome's performance in predicting gene expression and chromatin accessibility from personal genomes, finding it excels in accessibility but struggles with gene expression.

GAME: Genomic API for Model Evaluation

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

Deep genomic models of allele-specific measurements

This paper introduces DeepAllele, a deep learning model for allele-specific gene regulation analysis, showing improved detection of cis-regulatory elements in F1 hybrid mice immune cells.

Investigating Data Size, Sequence Diversity, and Model Complexity in MPRA-based Sequence-to-Function Prediction

The paper creates a curated MPRA dataset collection and builds sequence-to-function models to analyze how data size, diversity, and model complexity affect predictive performance.

Evaluation and optimization of sequence-based gene regulatory deep learning models

The paper evaluates and optimizes sequence-based gene regulatory deep learning models using a DREAM Challenge dataset, showing improvements through architectural and training strategy refinements.

A Supervised Contrastive Framework for Learning Disentangled Representations of Cell Perturbation Data

This paper introduces SC-VAE, a supervised contrastive framework for analyzing cell perturbation data, showing improved sensitivity in detecting perturbation effects compared to existing methods.

Spatial-linked alignment tool (SLAT) for aligning heterogenous slices properly

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

Identifying complex motifs in massive omics data with a variable-convolutional layer in deep neural network

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

Expectation pooling: An effective and interpretable pooling method for predicting DNA-protein binding

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

An exact transformation for CNN kernel enables accurate sequence motif identification and leads to a potentially full probabilistic interpretation of CNN

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