| Summary: | Cancer is a complex disease that involves genetic mutations and diverse tumour behaviour and
characteristics. With its complexities, there comes major challenges when it comes to treating
cancer such as requiring personalised treatment. Therefore, it is important for medical experts
to have a detailed understanding of patients’ cancer cells to be able to administer medicinal
efforts effectively.
In this day and age there is an abundance of data which also includes the various omics data of
cancer cells. With these omics data and integrating them together, medical experts can analyse
the relationships between each omics and obtain new insights into each biological component
during stages of cancer. This can help in understanding cancer cells as well as improving the
personalised treatment of cancer.
In this project, our end goal was to predict drug responses of cancer cell lines from multi-omics
data. However, multi-omics data has high dimensions which makes it difficult for integration
and analysis. Hence the approach we have taken to tackle this high dimensionality issue was
by implementing a dimension reduction technique using Variational Autoencoders (VAE).
Various integration techniques were also explored. Afterwards, a deep neural network predictor
was built to predict drug responses of cancer cells.
With this predictor, this will help in future drug and cancer research as well as improve current
cancer treatment.
|
|---|