Improving design, execution and analysis of transcriptomics experimentation

Over the last decades several high-throughput omics techniques have been developed with the promise and potential to greatly expand our knowledge about biology. Unfortunately, both of which have not fully been fulfilled to date. Here, several issues that could potentially cause this, were researched via use-case studies to identify which elements of the experimental chain of designing, executing and analyzing transcriptomics experimentation should be amended to enhance its reliability.
As starting point, we used our own experiences with a large-scale in-vitro study on the effects of UV-radiation on p53 in mouse embryonic fibroblasts (MEFs), that yielded only non-specific responses. For optimizing the design we looked into the use of small-scale range-finding prior to running a full-scale experiment to find the optimal locations in the design space to study specific responses of a process of interest. The execution was optimized by adjusting growth conditions of in-vitro MEFs to remove non-relevant stresses. In-vivo we developed a protocol for sampling biopsies from skin of mice and humans and subsequent isolation of totalRNA from these samples. To improve the analysis we looked into the quite substantial influence of so-called confounding factors on the outcome of experiments.
During the course of this research, some valuable lessons were learned about transcriptomics experimentation. Although many issues remain, over the last years, we and others have made quite some improvement to the design, execution and analysis of transcriptomics experiments, which will undoubtedly lead to better experiments, improved results and ultimately deeper insight in the complex biological systems.