CE42 - Capteurs, instrumentation

Automation of mechanobiological measurements by AFM, and their analysis by automatic learning – AutoBioTip

Submission summary

Autobiotip aims to awaken the AFM-Bio to large samples. We propose a method for automating force measurements to generate data on at least 1000 cells. The hypothesis is that this will allow us to access the heterogeneity of the mechanobiological properties of a population of cells. First, we will use directed assembly of cells and develop a process for automatic measurements. Then, traditional biophysical analyses and automatic learning (AI) methods will be adapted to extract the mechanobiological information from the large amount of data. The measurement of 1000 cells will be a breakthrough in the field (20 cells analysed in recent publications), and the use of AI to classify force curves is a new proposal. LAAS-CNRS (AFM-BIO, Biophysics and AI), ITAV-CNRS (AFM-Bio), labcom Biosoft (cell patterning) and IPN-CIC (automation) bring together the necessary skills to carry out the project.

The final products of this project will be of three types. At the instrumental level, we will develop cell immobilization substrates adapted to high throughput mechanobiology as well as a state-of-the-art autopilot system with rapid acquisition of a stack of force curves. In terms of data processing, we will develop a method of categorization and analysis by automatic learning that is totally original in the field. In the end, a platform for automatic analysis of the mechanical properties of large cell populations will be set up and made available to the biomedical community.

The main results will be: i) the demonstration of the possibility to measure in a reliable and reproducible way the mechanical properties of thousands of living cells in a few hours, ii) the demonstration that this type of analysis on a large population reveals new cellular behaviors so far masked by analyses on small numbers, iii) the demonstration that an analysis of stacks of force curves by automatic learning reveals relevant biological information.

Our approach will be divided into 3 parts. The first one gathers the methods necessary to mechanically probe a population of 1000 cells which are, the assembly of cells on known patterns and of controlled geometry, the development of a process for automated AFM measurements and the optimization of the sensitivity of the method. The second part is dedicated to the analysis of the large amount of data produced, and will be divided into 4 subparts. Firstly, the development of processes allowing the analysis of large quantities of data by the usual AFM data processing software, then by using supervised and unsupervised machine learning. Finally, we will carry out a step of comparison of these methods of analysis in order to point out the concordances and understand the divergences. These divergences will become potential sources of biophysical data not currently considered.

Project coordination

Etienne DAGUE (Laboratoire d'analyse et d'architecture des systèmes du CNRS)

The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.


RESTORE, a geroscience and rejuvenation research center
Institut Polytechnique National (IPN) / Centre d'Investigation en Informatique (CIC)
LAAS-CNRS Laboratoire d'analyse et d'architecture des systèmes du CNRS

Help of the ANR 326,496 euros
Beginning and duration of the scientific project: January 2021 - 48 Months

Useful links

Explorez notre base de projets financés



ANR makes available its datasets on funded projects, click here to find more.

Sign up for the latest news:
Subscribe to our newsletter