Building epigenetic clocks for estimating aging in life after cancer
“The epigenetic clock relies on the body’s epigenome, which comprises chemical modifications, such as methyl groups, that tag DNA. The pattern of these tags changes during the course of life, and tracks a person’s biological age, which can lag behind or exceed chronological age” (Nature News, 5.09.2019). Accelerating aging may be caused by environmental factors, including potentially cancer therapies. Scientists has constructed epigenetic clocks by selecting sets of DNA-methylation sites across the genome. There are few epigenetic clocks developed and their accuracy varies.
In many tumours, including testicular cancer (TC), a drug called cisplatin is given to patients to combat the cancer. The residues of cisplatin remain in vital human organs for many years after the treatment and are released to the blood stream throughout lifetime. Our group has recently found that cisplatin is altering DNA methylation pattern of patients who received this drug and data are available for further exploration.
The overarching aim of this study is to explore existing computational methods and develop new methods for estimating epigenetics clocks. Specifically,
- to compare normalization methods and performance of different existing epigenetic-clock
- to translate epigenetic clocks from methylation array platforms (tabular data) to whole genome bisulfite sequencing data (sequencing reads) by utilizing machine learning.
- to test if cisplatin causes accelerating aging in patients who were subjected to cisplatin, compared to those patients who survived TC without having cisplatin treatment.
The candidate will spend her/his time at the University of Oslo and Cancer Registry of Norway during the master project. No biological prior-knowledge is necessary. The candidate will be offered a special curriculum introductory course on cancer biology to be better prepared for the bioinformatic analyses in a biological context. Prior-knowledge in R or Python and an interest for machine learning methods is expected.
Supervisors: Marcin Wojewodzic, Trine B. Rounge