2021 Fiscal Year Research-status Report
Stem cell differentiation platform utilizing Bayesian machine learning
| Project/Area Number |
20KK0160
|
|---|
| Research Institution | Kyoto University |
|---|
Principal Investigator |
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
安井 孝介 京都大学, 高等研究院, 特定助教 (10877640)
ABDALKADER Rodi 京都大学, 高等研究院, 特定助教 (20839964)
|
|---|
| Project Period (FY) |
2020-10-27 – 2024-03-31
|
| Keywords | stem cell / cardiomyocite / machine learning / statistics / small molecule |
|---|
| Outline of Annual Research Achievements |
In order to generate tissues and organs in vitro for clinical or research purposes, cells of a specific type must be reproducibly and efficiently generated. Typically, such cells are generated by adding bioactive factors and nutrients to a stem cell culture medium. Unfortunately, it is not always clear which nutrients should be used to promote differentiation of stem cells into specific types of tissue. The goal of this project is to use machine learning and statistical methods to build a computational model which can predict nutrients for generating cardiac tissue from stem cells.
【FY2022 Achievement: establishment of a simple predictive model for small molecule nutrients on the basis of molecule shape and hydrophilicity information】
This predictive model was trained using a small dataset of (around 80 compounds) provided by research collaborators. Despite the model simplicity and small data set size, the model can predict the cardiomyocite differentiation effects of small molecules with over 80 % accuracy. The model incorporates shape information (based on the Zernike moment technique) and hydrophilicity information (based on molecular dynamics simulations), which appear to be the major factors determining the physiological activity of the molecules in stem cells.
|
| Current Status of Research Progress |
Current Status of Research Progress
2: Research has progressed on the whole more than it was originally planned.
Reason
The first of this project (the "development phase") was scheduled to run from October 2020 until August 2022. The computational model has developed according to plan, thanks to data provided by domestic collaborators as well as quick identification of the molecular features (shape and hydrophilicity) responsible for cardiac differentiation. A paper on this model could easily be written, however we wish to acquire further data first in order to raise the impact of our study. On the other hand, our data collection plans are delayed because we have been unable to travel to the United States for the relevant experiments (including single-cell RNA sequencing experiments) due to COVID-19. We hope to travel to the United States during FY2022 in order to proceed with this part of the research.
|
| Strategy for Future Research Activity |
[Data acquisition for further model validation, April - September 2022]
A technician will join this project in the early summer, who will collect further data for training and testing the computational model. These results will be used to improve the model, either by enabling better parameter estimates or by identifying types of molecules for which the model performs poorly. The model will be improved as appropriate.
[Prediction and synthesis of new molecules, August 2022 - March 2023]
After validating the computation model further, we will predict new molecules with strong cardiac differentation effects, have them synthesised, and test them in the laboratory. We also wish to acquire single-cell RNA sequencing data in order to understand the precise effects of those molecules on stem cells.
|
| Causes of Carryover |
・Travel plans were unexpected delayed due to the pandemic. Therefore, we wish to use the FY2021 travel budget in FY2022.
・There were plans to employ a research technician in FY2021. This technician was to be sent overseas to collect data. However, due to travel restrictions, we decided not to employ a technician in FY2021. We therefore wish to use the FY2021 personnel budget in FY2022 instead.
|
