PreprintCC BY-NCbioRxiv
DELPHAI predicts heterogeneous perturbation responses with learned single-cell fitness
- 1aiPTO TechBio
Abstract
Current perturbation response modelling in single-cell transcriptomics assumes conserved cell mass and loses gene expression information to latent-space decoding. We propose DELPHAI, training a fitness network and an optimal transport network jointly without biological priors, and during inference applying a fitness-gated transport with a direct gene-space retrieval. Demonstrated across two benchmark frameworks, DELPHAI ranks first in predicting differentially expressed genes, while revealing which cell lineages a perturbation depletes.
Related
- Mitochondrial perturbation adapts the proteome from early to advanced immune responses
- Emotional arousal predicts pupil responses following a single attention shift between bright and dark moving stimuli
- Long-term biodiversity responses to anthropogenic change are spatially and taxonomically heterogeneous across coastal wetlands
- Multidimensional profiling of heterogeneous lateral habenula subpopulations reveals distinct responses during motivated behavior
- Single-Cell Proteomics Reveals Heterogeneous and Bimodal Proteome Responses to DNA Damage
BibTeX
@misc{zhang2026delphai,
title = {{DELPHAI predicts heterogeneous perturbation responses with learned single-cell fitness}},
author = {X. Zhang and H. Wu and H. Liu},
howpublished = {bioRxiv},
year = {2026},
month = jul,
doi = {10.64898/2026.07.01.735965},
url = {https://www.biorxiv.org/content/10.64898/2026.07.01.735965v2},
note = {CC BY-NC},
}
RIS
TY - UNPB
TI - DELPHAI predicts heterogeneous perturbation responses with learned single-cell fitness
AU - X. Zhang
AU - H. Wu
AU - H. Liu
PY - 2026
DA - 2026/07/01
PB - bioRxiv
DO - 10.64898/2026.07.01.735965
UR - https://www.biorxiv.org/content/10.64898/2026.07.01.735965v2
AB - Current perturbation response modelling in single-cell transcriptomics assumes conserved cell mass and loses gene expression information to latent-space decoding. We propose DELPHAI, training a fitness network and an optimal transport network jointly without biological priors, and during inference applying a fitness-gated transport with a direct gene-space retrieval. Demonstrated across two benchmark frameworks, DELPHAI ranks first in predicting differentially expressed genes, while revealing which cell lineages a perturbation depletes.
ER -