guest ::
| Home > Publications database > Point cloud transformers for parameter inference of large-scale tissue simulations |
| Home > Publications database > Point cloud transformers for parameter inference of large-scale tissue simulations |
| Abstract | FZJ-2025-01062 |
; ;
2024
This record in other databases:
Please use a persistent id in citations: doi:10.1016/j.bpj.2023.11.1988
Abstract: Computational modeling plays a pivotal role in unraveling the intricate dynamics of living tissues, such as tumor growth. Yet, the task of deriving quantitatively meaningful parameters from experimental data often poses a challenge. Traditional methods like approximate Bayesian computation rely on summary statistics to gauge the congruence between simulations and real-world experiments, an approach with inherent limitations due to the difficulty of selecting these statistics. To surmount these issues, we advocate for the adoption of simulation-based inference (SBI), harnessing the power of deep learning techniques to navigate these complexities. In this work we elucidate how point cloud transformers can be employed directly on the positional data of in-vitro spheroids for parameter inference, circumventing the need of summary statistics. We integrate the training of these neural networks into the parameter inference pipeline of cells in silico (CiS), our high-performance framework designed for large-scale tissue simulations. This novel approach not only yields superior results in terms of inference accuracy but also significantly enhances computational efficiency compared to traditional methodologies. By enabling the use of parameters that more faithfully capture experimental data, our approach empowers researchers to explore pressing biological questions. For instance, we can investigate the intricate interplay between the extracellular matrix and tumor invasion behavior, shedding new light on critical aspects of cancer biology.
; BIOSIS Previews ; Biological Abstracts ; Clarivate Analytics Master Journal List ; Current Contents - Life Sciences ; Essential Science Indicators ; SCOPUS ; Science Citation Index Expanded ; Web of Science Core Collection
|
The record appears in these collections: |
PDF