Scientists from Tokyo Metropolitan University used machine learning to automate the identification of defects in sister chromatid cohesion. They trained a convolutional neural network (CNN) using micrographs of individual stained chromosomes, which the researchers identified as having or without coherence defects. After the training, I was able to successfully classify 73.1% of the new images. Automation promises better statistics and more insight into the wide range of perturbations that cause coherence defects.
Chromosomes are made up of long DNA molecules that contain part of our genes. When cells divide, the chromosomes must be copied so that both new cells contain all the information they need to function. This is accomplished through DNA replication, creating two identical copies known as sister chromatids which are held together by a ring-like protein structure called a cohesin. It is necessary to keep these copies together during cell division. Cohesion problems can lead to the breakdown of chromatids, causing serious disruption to the healthy functioning of cells and organs.
The study of cohesion defects in chromosomes has been largely done by researchers observing chromosomes under a microscope. With a special dye, experienced scientists can tell whether or not the chromatids are linked in the right way. This type of classification is vital in the study of chromosomal defects, including the correct functioning of cohesion. However, the entire process is manual. When statistics are needed on how many chromosomes are in the correct or incorrect state, the process becomes very inefficient, taking a large number of man-hours by experienced scientists.
Now, an interdisciplinary team of biologists and machine learning specialists from Tokyo Metropolitan University led by Associate Professor Takuya Abe, Professor Kiyoshi Nishikawa, Associate Professor Kan Okubo, and Professor Koji Hirota have joined forces to automate this time-consuming process. They used the same technology that powers facial recognition and machine vision to analyze microscopic images of chromosomes with and without cohesion defects. They used a convolutional neural network (CNN), a type of machine learning algorithm particularly suited to image recognition, and trained it on more than 600 images of chromosomes that had been previously categorized into three groups manually by the scientists. By the end of the process, new images fed through the algorithm can be classified in the same way as experienced researchers with 73.1% accuracy. This has the potential to greatly simplify and speed up chromosome experiments.
The team also used a cell line that removed a gene known to affect cohesion called CTF18 Chromosomes were analyzed using a trained neural network. The network found significant differences between normal and normal cells CTF18 disrupted cells, suggesting that the network, on its own, was able to pick up on genetic problems that affected coherence. Although their method currently only recognizes three groups, it can be expanded to include different haplotypes in different species, enabling rapid classification and unprecedentedly accurate quantification of chromosomal defects in a wide range of diseases.
This work was supported by the Uehara Memorial Foundation, Mochida Memorial Foundation for Medical and Pharmaceutical Research, Kanae Foundation for the Promotion of Medical Sciences, Senri Life Science Foundation, JSPS KAKENHI Grant Numbers 17K17986, 20K06760, 22H05072, 22K12170, 20H04337 and 19 KK0210. 16 H01314.
Ikemoto, D.; et al. (2023) Application of neural network-based image analysis to detect sister chromatid cohesion defects. Scientific reports. doi.org/10.1038/s41598-023-28742-6.