TY - JOUR
T1 - Predicting cellular responses to complex perturbations in high-throughput screens
AU - Lotfollahi, Mohammad
AU - Klimovskaia Susmelj, Anna
AU - De Donno, Carlo
AU - Hetzel, Leon
AU - Ji, Yuge
AU - Ibarra, Ignacio L.
AU - Srivatsan, Sanjay R.
AU - Naghipourfar, Mohsen
AU - Daza, Riza M.
AU - Martin, Beth
AU - Shendure, Jay
AU - McFaline-Figueroa, Jose L.
AU - Boyeau, Pierre
AU - Wolf, F. Alexander
AU - Yakubova, Nafissa
AU - Günnemann, Stephan
AU - Trapnell, Cole
AU - Lopez-Paz, David
AU - Theis, Fabian J.
N1 - Publisher Copyright:
© 2023 The Authors. Published under the terms of the CC BY 4.0 license.
PY - 2023/6/12
Y1 - 2023/6/12
N2 - Recent advances in multiplexed single-cell transcriptomics experiments facilitate the high-throughput study of drug and genetic perturbations. However, an exhaustive exploration of the combinatorial perturbation space is experimentally unfeasible. Therefore, computational methods are needed to predict, interpret, and prioritize perturbations. Here, we present the compositional perturbation autoencoder (CPA), which combines the interpretability of linear models with the flexibility of deep-learning approaches for single-cell response modeling. CPA learns to in silico predict transcriptional perturbation response at the single-cell level for unseen dosages, cell types, time points, and species. Using newly generated single-cell drug combination data, we validate that CPA can predict unseen drug combinations while outperforming baseline models. Additionally, the architecture's modularity enables incorporating the chemical representation of the drugs, allowing the prediction of cellular response to completely unseen drugs. Furthermore, CPA is also applicable to genetic combinatorial screens. We demonstrate this by imputing in silico 5,329 missing combinations (97.6% of all possibilities) in a single-cell Perturb-seq experiment with diverse genetic interactions. We envision CPA will facilitate efficient experimental design and hypothesis generation by enabling in silico response prediction at the single-cell level and thus accelerate therapeutic applications using single-cell technologies.
AB - Recent advances in multiplexed single-cell transcriptomics experiments facilitate the high-throughput study of drug and genetic perturbations. However, an exhaustive exploration of the combinatorial perturbation space is experimentally unfeasible. Therefore, computational methods are needed to predict, interpret, and prioritize perturbations. Here, we present the compositional perturbation autoencoder (CPA), which combines the interpretability of linear models with the flexibility of deep-learning approaches for single-cell response modeling. CPA learns to in silico predict transcriptional perturbation response at the single-cell level for unseen dosages, cell types, time points, and species. Using newly generated single-cell drug combination data, we validate that CPA can predict unseen drug combinations while outperforming baseline models. Additionally, the architecture's modularity enables incorporating the chemical representation of the drugs, allowing the prediction of cellular response to completely unseen drugs. Furthermore, CPA is also applicable to genetic combinatorial screens. We demonstrate this by imputing in silico 5,329 missing combinations (97.6% of all possibilities) in a single-cell Perturb-seq experiment with diverse genetic interactions. We envision CPA will facilitate efficient experimental design and hypothesis generation by enabling in silico response prediction at the single-cell level and thus accelerate therapeutic applications using single-cell technologies.
KW - generative modeling
KW - high-throughput screening
KW - machine learning
KW - perturbation prediction
KW - single-cell transcriptomics
UR - http://www.scopus.com/inward/record.url?scp=85158160308&partnerID=8YFLogxK
U2 - 10.15252/msb.202211517
DO - 10.15252/msb.202211517
M3 - Article
AN - SCOPUS:85158160308
SN - 1744-4292
VL - 19
JO - Molecular Systems Biology
JF - Molecular Systems Biology
IS - 6
M1 - e11517
ER -