Automating a Yeast Reverse Genetic Screen
The client was interested in better understanding the complex regulatory network controlling cell division in yeast. Since no large-scale genetic screen had been published, the amount of data available was limited. Existing datasets published by various groups were based on different yeast strains and methods. This made it very difficult to compare data produced by different groups and develop a good understanding of the regulatory network.
Since third-party data was too sparse for reliable analysis, they decided to perform a high-throughput reverse genetic screen. They characterized the genetic interactions between the 36 genes involved in the control of the cell cycle in yeast. Using synthetic genetic array technology, they produced all the 1,296 double knock-out mutants. This experimental design would help them detect genes that work together to control cell division.
While the client has expertise in standard yeast genetics and molecular biology techniques, they were struggling to scale up their processes. Their instruments were automated, so they believed they had the resources they needed to pull it off. But in their first attempt to increase the scale of their experiments, they discovered they needed to automate their workflows as well.
They spent two years on this first attempt to make and test this yeast mutant library. They were producing more data than they could manage, let alone analyze.
Eventually, they had to stop laboratory operations to take the time to look at the data they had spent so much time and effort collecting. They reached out to GenoFAB to help them organize and analyze their data. Data analysis revealed that the experiment relied on a key hypothesis that proved incorrect.
This meant that the experiment had to be done again as none of the data collected during the first two years was usable.
GenoFAB helped the client automate key aspects of this large scale yeast genetic screen. Before doing the experiment again, we helped the client define the entire process. It was essential to get the big picture, from the generation of yeast mutants all the way to the final analysis of the phenotypic data. This global perspective on the project helped us develop the data model needed to capture data at key stages of the process.
This comprehensive approach allowed us to develop project dashboards to keep track of each mutant progression throughout the project life.
Design: selection cassettes and PCR primers
We designed 180 selection cassettes by combining 36 homologous recombination sequences and 5 selection cassettes for each of the 36 genes. We also designed 360 primers to generate 180 DNA fragments by amplifying selection cassettes from existing plasmids. Finally, we leveraged our automated bioinformatics scripts to streamline the design of PCR primers to verify parent lines.
Build: production of the strain collection
We helped the client design yeast transformation and yeast mating workflows. We optimized process stability and reproducibility prior to scaling up to minimize the cost of failed experiments. Another contribution was the optimization of the layout of microtiter plate designs to facilitate automation. In order to produce more than 5000 crosses, it was important to limit steps that would have slowed down the process and created opportunities for errors.
All the workflows were designed to allow technicians with little prior molecular biology or yeast genetics experience to complete specific tasks reliably. The overall process was also designed to maximize data quality while avoiding unmanageable workloads.
We supported the generation of sample labels using barcode technologies. Success of the experiment depended on the ability to track thousands of samples at all steps of the process.
Test: high-throughput phenotyping
We also helped the client automated the phenotyping of the yeast mutant collections. Fitness data were collected using colony size assays. The process required them to test more than 6,000 strains at six different time points on six different media.
We helped the team design plates to ensure that data could be analyzed. It was very important to have the right controls at the right locations on the plate to be able to extract information from the raw data.
We helped manage phenotypic data by designing a custom database to import colony size data produced by the plate imager and connect the phenotypic data and the samples recorded in the LIMS.
We developed data analysis services to perform the statistical analysis of colony size. These scripts included spatial normalization based on the distance of individual colonies from the plate edge.
Learn: analysis of phenotypic data
We developed data analysis scripts to estimate the growth rate of colonies of individual mutants. We compared these estimates to the growth rate of the control strain after filtering out outlier colonies resulting from pinning errors. We derived estimates of fitness from growth rates and estimates of genetic interactions.
This high-throughput yeast genetic screen uncovered major reproducibility issues with the qualitative test that was the commonly accepted standard in the field. Through the help and guidance of GenoFAB, the client was able to produce quantitative datasets that are more reproducible than the standard qualitative data. They have produced the most comprehensive set of data available to their community of interest. Analysis of this new kind of data is ongoing but it is expected to become the new standard in the field.