MuSe-GNN: Learning Unified Gene Representation From Multimodal Biological Graph Data

Authors: Tianyu Liu, Yuge Wang, Rex Ying, Hongyu Zhao

NeurIPS 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Comprehensive benchmarking analysis shows our model s capacity to effectively capture gene function similarity across multiple modalities, outperforming state-of-the-art methods in gene representation learning by up to 97.5%. Moreover, we employ bioinformatics tools in conjunction with gene representations to uncover pathway enrichment, regulation causal networks, and functions of disease-associated or dosage-sensitive genes.
Researcher Affiliation Academia Tianyu Liu Yale University tianyu.liu@yale.edu Yuge Wang Yale University yuge.wang@yale.edu Rex Ying Yale University rex.ying@yale.edu Hongyu Zhao* Yale University hongyu.zhao@yale.edu
Pseudocode Yes Algorithm 1 Multimodal Similarity Learning Graph Neural Network (Mu Se-GNN)
Open Source Code Yes 1Codes of Mu Se-GNN: https://github.com/Hello World LTY/Mu Se-GNN
Open Datasets Yes Leveraging 82 training datasets from 10 tissues, three sequencing techniques, and three species, we create informative graph structures for model training and gene representations generation... Download links: Appendix M. ... We used Mu Se-GNN to generate gene embeddings for different datasets based on an unsupervised learning framework and utilized the gene embeddings as training dataset to predict the function of genes based on k-NN classifier.
Dataset Splits No The paper does not explicitly state training/validation/test dataset splits with specific percentages or counts for reproducibility. It discusses evaluation metrics and benchmarking but not a dedicated validation split.
Hardware Specification No The paper does not explicitly describe the specific hardware used to run the experiments (e.g., specific GPU/CPU models, memory, or cloud resources).
Software Dependencies No The paper mentions software like "Scanpy [104]" and tools like "ggplot2 [102]" but does not provide specific version numbers for these or other key software dependencies required to reproduce the experiments.
Experiment Setup Yes Details of hyper-parameter tuning can be found in Appendix E.2.