Causal single-cell RNA-seq simulation, in silico perturbation, and GRN inference benchmarking using GRouNdGAN-Toolkit
Bibliographic record
Abstract
Abstract Background Rapid advances in high-throughput single-cell sequencing technologies, coupled with the development of computational methods capable of leveraging large datasets, have led to the emergence of numerous approaches for deciphering regulatory interactions in the form of Gene Regulatory Networks (GRNs). However, in the absence of context-specific gold-standard ground truths, particularly those containing causal interactions, systematically benchmarking GRN inference methods remains a challenge. Thus, we previously developed GRouNdGAN, a causal implicit generative model capable of simulating realistic observational and interventional scRNA-seq data following a user-defined GRN on any biological system of interest. Importantly, we demonstrated that GRouNdGAN generates datasets that bridge the gap between experimentation and simulation for GRN inference benchmarking. Method Building upon the GRouNdGAN framework, we developed an extended toolkit that offers additional features, including interactive model visualization, training monitoring, more customizable GRN creation options, synthetic data similarity and GRN inference benchmarking metrics, and an intuitive TF knockout prediction module. Here, we provide a step-by-step procedure for implementing the protocol from start to finish and introduce alternative variations to adapt GRouNdGAN to studies with different experimental setups. GRouNdGAN-Toolkit is publicly available as a python code repository and containerized application and is accompanied by a tutorial website featuring a collection of simulated datasets. Model training largely depends on graphic hardware and the size and density of the input GRN, and typically takes around 75h to complete on a single GPU. Excluding model training, this protocol typically takes less than 25min to complete. Discussion GRouNdGAN-Toolkit is a versatile simulator with user friendly interface that does not assume advanced computational genomics expertise, enhancing its usability and accessibility across a wide range of users.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".