One of the most critical and under-appreciated activities in science is “hit-picking.” Hit-picking is like the Kentucky Derby of everyday science. It’s when you line up the results of weeks or months of bioengineering work (strains, tissues, enzymes) and pick the winners that will go on to further development and testing. Get this right and you may have a billion dollar drug or a ground-breaking scientific publication. Get this wrong and you may just spend the next 12 months of your life going in circles of frustration — never making progress.
All-too-often, the hit-picking process suffers from a few correctable issues. Sometimes the statistical rigor is simply lacking (e.g., the target false positive or false negative rates are not determined or adhered to); sometimes data is not adequately controlled for week-to-week variations in assay processes which leads to biased selections; and sometimes choices are made based on anecdotal information not apparent in the data records. Regardless, nearly all hit-picking is done without adequate capture of the rationale, even for the most rigorously data-driven selections. The consequence is that you have no historical record to judge and improve your strains or your selection process.
The cost of sub-optimal hit-picking processes can be substantial, both to those creating the candidates and those responsible for downstream development. A false-negative can throw away months of hard work by the teams selecting the candidates. A false positive can cost months of misplaced effort for the downstream teams. The hit-picking process deserves as much tender love and care as the laboratory work that precedes or follows it.
This is one of deepest and most fundamental motivators for the design of Riffyn Nexus. Riffyn Nexus allows you to find and pick hits graphically, statistically, with formulas or just plain old sorting (if you wish). Once you pick your hits you can memorialize such analytical decisions by saving your selections and selection reasons directly onto the source data for both immediate action on running experiments, or for future analysis. And like everything in Riffyn, your data is linked to its source. So when you mark your hit-picking decisions on the final sample measurements, you can also tag all the source strains, reagents and materials that generated those final samples.
We make it possible to perform these selection decisions directly in the Riffyn Nexus, or via external visualization and analytics tools where you can plot, analyze and interactively select your hits. Either way, you can then save your selections automatically to your experiment source data in Riffyn Nexus. You can even couple Riffyn Nexus to automated data analytics or machine learning pipelines which run in real-time in the background, and then save their picks directly to the original experiment side by side with any manually selected picks.
For JMP in particular, we have built an AddIn that lets you visually select data and save tags on those selections back to Riffyn automatically. The 5 minute video at the top of this post is an example of just that.
If you’d like to give it a try, just give us a shout: firstname.lastname@example.org.
Check out Riffyn's Discover Blog to learn more about optimal experimental design and analysis, such as an easier way to design DoEs, how to optimize your measurement systems, how to structure data for analysis, or to clarify what machine learning is and how to apply it to your systems.
Fun Fact: In addition to Riffyn's JMP add-in, Riffyn also has a Data Connector for Tableau!
Process and experiment design were created, and data were captured and prepared with Riffyn.
Plots were created with JMP® Software. Copyright 2018, SAS Institute Inc., Cary, NC, USA. All Rights Reserved. Reproduced with permission of SAS Institute Inc., Cary, NC.