Design-of-Experiments for Nonlinear, Multivariate Biology: Rethinking Experimental Design through Perturb-seq
Okano, Y.; Ishikawa, T.; Sato, Y.; Okano, H.; Sakurada, K.
Show abstract
Design of experiments (DOE) principles are increasingly applied to biological assays, yet it remains unclear whether the optimality of their foundational assumption--orthogonal decomposition--holds in nonlinear biological systems. We addressed this question using Perturb-seq as a case study. By benchmarking a design commonly used in Perturb-seq and related experiments against the orthogonal Plackett-Burman (PB) design via simulations, we uncovered a counter-intuitive phenomenon: while orthogonal designs generally excel, the correlated structure inherent to the common design is functionally robust in systems with significant signal amplification. This challenges the blind application of DOE to biology. Based on these findings, we developed the PB suitability index (PBSI), a simple, parameter-free metric that predicts the optimal design solely from network structure. Our work not only provides practical guidelines for Perturb-seq but also establishes a "biology-oriented DOE" framework, bridging the gap between statistical rigor and biological complexity.
Matching journals
The top 4 journals account for 50% of the predicted probability mass.
Similar papers in this journal
Similar papers in this journal
- A general model-based causal inference overcomes the curse of synchrony and indirect effect 96%
- FastCCC: A permutation-free framework for scalable, robust, and reference-based cell-cell communication analysis in single cell transcriptomics studies 96%
- Increasing certainty in systems biology models using Bayesian multimodel inference 95%
Similar papers in this journal
- DeepDRIM: a deep neural network to reconstruct cell-type-specific gene regulatory network using single-cell RNA-seq data 94%
- Predicting 3D genome architecture directly from the nucleotide sequence with DNA-DDA 94%
- SMNN: Batch Effect Correction for Single-cell RNA-seq data via Supervised Mutual Nearest Neighbor Detection 94%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.