Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..

Learning Crossmodal Interaction Patterns via Attributed Bipartite Graphs for Single-Cell Omics

Authors: Xiaotang Wang, Xuanwei Lin, Yun Zhu, Hao Li, Yongqi Zhang

NeurIPS 2025 | Venue PDF | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments demonstrate that Bi2Former achieves state-of-the-art performance in crossmodal matching across diverse datasets, remains robust under sparse training data, generalizes to unseen cell types and datasets, and reveals biologically meaningful regulatory patterns.
Researcher Affiliation Academia Xiaotang Wang The Hong Kong University of Science and Technology (Guangzhou) EMAIL Xuanwei Lin Fuzhou University EMAIL Yun Zhu Shanghai Artificial Intelligence Laboratory EMAIL Hao Li Academy of Military Medical Sciences EMAIL Yongqi Zhang The Hong Kong University of Science and Technology (Guangzhou) EMAIL
Pseudocode No The paper describes the model architecture and mathematical equations for its components but does not include a clearly labeled pseudocode or algorithm block.
Open Source Code Yes Our code is available at: https://github.com/wangxiaotang0906/Bi2Former.
Open Datasets Yes To ensure the reliability and comparability of our evaluation, we conduct experiments on five widely-used benchmark datasets for single-cell omics: ISSAAC-seq [50], 10 Multiome PBMC [1], SHARE-seq [28], SNARE-seq [8], and 10 genomics Multiome.
Dataset Splits Yes As summarized in Table 5, we maintain a 1:1 ratio of positive to negative pairs, resulting in a graph dataset that contains twice the number of samples as the original single-cell dataset. ... To assess the generalization capability of our method, we evaluate the performance of Bi2Former under a cross-cell-type setting. We split each dataset into training and test sets with disjoint cell types in a 1:1 ratio (See details in Appendix B.3). ... Table 5: Statistics of our ABG datasets. ... Split(%) 60/20/20
Hardware Specification No The paper does not provide specific hardware details (e.g., GPU/CPU models, memory) used for running the experiments.
Software Dependencies No The paper does not provide specific software dependencies with version numbers.
Experiment Setup Yes Other settings. We report experimental results using hyperparameter settings detailed in Appendix B.4, selecting those that achieve the highest validation performance. ... learning_rate {1e 3, 5e 4, 1e 4, 5e 5, 1e 5} weight_decay {1e 4, 5e 5, 1e 5, 5e 6, 1e 6} dropout {0, 0.1, 0.3, 0.5, 0.8} For Bi2Former, ID embedding dims {64, 128, 256, 512} hidden dims {64, 128, 256, 512} layer_num {1, 2}