Скачать или смотреть Bayesian computational analysis of cell division dynamics

Talk by Constandina Koki at BioInference 2023.

https://bioinference.github.io/

Abstract: Millions of cell divisions take place in our body on a daily basis, divisions that have to accurately replicate the cell’s chromosomes. In particular newly duplicated chromosomes – called sister chromatids – must be segregated into daughter cells with high fidelity. Errors in this process, or mis-segregation is a leading cause of cancer. Segregation is a complex mechanical process requiring sisters to cor- rectly attach to the bipolar mechanical spindle machine, thereby moving the chromosomes. Recent innovations in imaging allow 3D high resolution tracking of sister chromatids in living human cells throughout cell division (40 mins). We deploy Bayesian computational techniques to analyse these time-series by fitting biophysically motivated models to infer attachment states. We utilise and com- pare models with hidden discrete mechanical states and a continuum. Our analysis reveals that sister dynamics are significantly asymmetric in 20% of pairs, likely resulting from asymmetric attachments. To deal with the missing data (detection loss during tracking) we utilise a “reduced” likelihood that is based only on the actual data. Model inference is carried out in STAN.