Comprehensive View Embedding Learning for Single-Cell Multimodal Integration
Authors: Zhenchao Tang, Jiehui Huang, Guanxing Chen, Calvin Yu-Chian Chen
AAAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experimental results on multiple public datasets show that Co VEL is accurate and robust to single-cell multimodal integration. |
| Researcher Affiliation | Academia | Zhenchao Tang1, Jiehui Huang1, Guanxing Chen1, Calvin Yu-Chian Chen1,2,3,4,5* 1School of Intelligent Systems Engineering, Shenzhen Campus of Sun Yat-sen University 2AI for Science (AI4S) Preferred Program, Peking University Shenzhen Graduate School 3School of Electronic and Computer Engineering, Peking University Shenzhen Graduate School 4Department of Medical Research, China Medical University Hospital 5Department of Bioinformatics and Medical Engineering, Asia University |
| Pseudocode | No | The paper describes the proposed method using prose and mathematical equations but does not include any explicitly labeled 'Pseudocode' or 'Algorithm' blocks. |
| Open Source Code | Yes | Data availability: https://github.com/shapsider/scintegration. |
| Open Datasets | Yes | Mouse skin dataset (Ma et al. 2020): ... Mouse cortex dataset (Saunders et al. 2018; Luo et al. 2017): ... Chen-2019 used SNARE-seq technology to jointly measure 9,190 cells in mouse cortex (Chen, Lake, and Zhang 2019). 10x-Multiome used 10x-Multiome technology to jointly measure 9,631 cells of human PBMC (Cao and Gao 2022). Muto-2021 measured 44,190 cells from human kidney using sn RNA-seq and sn ATAC-seq technologies, respectively (Muto et al. 2021). Yao-2021 measured 124,571 cells from mouse MOp using 10x RNA v3 and sn ATAC-seq technology (Yao et al. 2021). |
| Dataset Splits | No | For multimodal integration task, all methods learn joint embedding unsupervisedly on the entire dataset. The paper mentions evaluating on 'test set' and 'subsampled datasets' but does not specify training, validation, or test dataset splits in terms of percentages or counts for model development and evaluation stages. |
| Hardware Specification | Yes | Co VEL trained using NVIDIA Ge Force RTX A6000 with 48 GB memory. |
| Software Dependencies | No | All datasets are preprocessed according to the standard of scanpy (Wolf, Angerer, and Theis 2018). The paper mentions the use of 'scanpy' but does not provide version numbers for any other software dependencies like deep learning frameworks or specific libraries. |
| Experiment Setup | Yes | Adam optimizer with 0.001 learning rate was used to update model parameters. The batch size was set to 16. ... Graph encoder is 2-layer GAT ... We set 2 layers Teacher Transformer. |