Large-Scale Differentiable Causal Discovery of Factor Graphs

Authors: Romain Lopez, Jan-Christian Huetter, Jonathan Pritchard, Aviv Regev

NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental DCD-FG uses a Gaussian non-linear low-rank structural equation model and shows significant improvements compared to state-of-the-art methods in both simulations as well as a recent large-scale single-cell RNA sequencing data set with hundreds of genetic interventions.
Researcher Affiliation Collaboration 1 Division of Research and Early Development, Genentech {lopez.romain, huettej1, regeva}@gene.com 2 Department of Genetics, Stanford University pritch@stanford.edu 3 Department of Biology, Stanford University
Pseudocode No The paper does not contain any sections or figures explicitly labeled as 'Pseudocode' or 'Algorithm'.
Open Source Code Yes 3. If you ran experiments... (a) Did you include the code, data, and instructions needed to reproduce the main experimental results (either in the supplemental material or as a URL)? [Yes] (Appendix)
Open Datasets Yes We focus on a recent Perturb-CITE-seq experiment [14] that contains expression profiles from 218,331 melanoma (cancer) cells, after interventions targeting each of 249 genes. and the citation: [14] Chris J Frangieh, Johannes C Melms, Pratiksha I Thakore, Kathryn R Geiger-Schuller, Patricia Ho, Adrienne M Luoma, Brian Cleary, Livnat Jerby-Arnon, Shruti Malu, Michael S Cuoco, et al. Multimodal pooled Perturb-CITE-seq screens in patient models define mechanisms of cancer immune evasion. Nature Genetics, 53(3):332 341, 2021.
Dataset Splits Yes For every model, we performed a hyperparameter search using a goodness of fit metric on a small validation set.
Hardware Specification Yes on one NVIDIA Tesla T4 GPU with 15Gb of RAM
Software Dependencies No The paper mentions 'Python 3.8' and packages like 'PyTorch and SciPy' and 'Scanpy [63]' but does not provide specific version numbers for the libraries/packages used.
Experiment Setup No We provide further details on all experiments, including the grids used for hyperparameter search, as well as supplementary experiments, in Appendix F.