Roth A, et al. Clonal genotype and population structure inference from single-cell tumor sequencing. Nature Methods. 2016 May 16. doi: 10.1038/nmeth.3867. [Epub ahead of print]
Our Nature Methods paper details a new statistical approach for analysing single cell genomics data. The manuscript describes a novel statistical model coupled with a mean field variational inference method to address the problems of: missing values, biased allelic counts, and false measurements of genotypes due to sequencing multiple cells which are common in single cell sequencing datasets.
Situated within the tech hub of Vancouver, B.C., Dr. Sohrab Shah’s bioinformatics lab at the BC Cancer Agency has developed a new machine learning tool enabling the study of the role of individual cancer cells in cancer progression. Published today in Nature Methods, Shah’s team shows the power of digital cancer biology through computational analysis of mutations in individual ovarian cancer cells.
Developed by Dr. Andrew Roth, the open source software called Single Cell Genotyper is a new statistical model and machine learning inference algorithm designed to determine the pattern of how DNA mutations are distributed in individual cells of a tumour. This provides unprecedented digital resolution to identify the number of different types of cancer cells present in a tumour and to track how they migrate when the disease spreads or relapses.
The Single Cell Genotyper model allowed Shah and his team to reveal critical insight into the invasive spread of the most malignant form of ovarian cancer. These findings are simultaneously published in a landmark study in Nature Genetics. This is a first in mapping two distinct patterns of cancer cell migration in the most deadly form of ovarian cancer.
“Cancer science is now a quantitative, digital science,” says Shah of their technological advances that are making big strides in advancing cancer knowledge.
“The Single Cell Genotyper software led us to define cell migration maps for the first time in ovarian cancer, including what cell types are present, where they are found in the abdominal cavity and their migration patterns from site to site,” he added.
Single Cell Genotyper
Measurements of mutations in individual cancer cells are input into the SCG which is able to work through the ‘noise’ or ‘interference’ of competing or partially missing data to efficiently:
- Estimate the number of cancer cell populations present in a tumour
- Identify the set of mutations that define each population
- Predict the abundance of each population in the tumour
This technology provides a new tool scientists can use to study the cell-population composition of all types of human cancer. This is a necessary first step to understand how cancers acquire resistance to treatment and spread beyond their site of origin.
Next steps are to apply the innovative Single Cell Genotyper to define cell migration maps in ovarian cancer and breast cancer patients with a specific focus on determining which cells are resistant to treatment and what are their specific properties. This will allow researchers to build predictive tools to better inform future care.
The research findings were made possible with philanthropic investments from the BC Cancer Foundation.
Notes: Dr. Sohrab Shah is a Senior Scientist at the BC Cancer Agency, Associate Professor at the University of British Columbia and Canada Research Chair in Computational Cancer Genomics. Dr. Andrew Roth is a postdoctoral fellow at the BC Cancer Agency and UBC.