Improved in vitro model selection for cancer using molecular data
Typ
Examensarbete för masterexamen
Program
Biotechnology (MPBIO), MSc
Publicerad
2020
Författare
Andersson, Emma Charlotte
Mac-Lean Ballivián, María Trinidad
Modellbyggare
Tidskriftstitel
ISSN
Volymtitel
Utgivare
Sammanfattning
Cancer cell lines are essential components in the process of cancer drug research due to their function as early stage models for primary tumors. However, selection of the optimal cell line model can be complicated and therefore many drug projects fail because the chosen cell line does not give sufficient results, whilst another model might. Furthermore, many of the candidate cell lines for drug studies are not always good models, only convenient because of their availability and high growth rate. This does, in many cases, result in waste of resources and incapability of drawing good conclusions within a research project. Thus, there is an apparent need for a smarter cell line selection.
In this project, a cell line selection workflow was carried out based on comparison of genomic copy number and transcriptomic data from cell lines and primary tumors. Tumor samples were grouped after tumor subtypes, to make the sub-populations more homogeneous, thereby enabling selection of cell lines resembling only tumor samples similar to the tumor phenotype of interest. The comparison was carried out, first using correlative analysis within copy numbers, then by expression correlation. Thereafter, the search for optimal cell line models was continued by investigating specific traits for the given primary tumor, using gene signature expression.
To demonstrate cell line selection using subtypes, we present an example; the breast cancer tumor subtype Her2 with DNA damage response deficiency. For this subtype, we found 10 cell lines among the top candidates from both the transcriptomic and the genomic correlative analysis, whereafter 5 were selected based the gene signature expression analysis of DNA damage response deficiency. The cell lines found as the best models for breast cancer Her2 tumors of this specific phenotype, were UACC893, KPL1, SW1990, BT474 and HCC1419. Most of these cell lines originate from breast cancer tissue, except for SW1990 which is derived from a pancreatic adenocarcinoma tumor.