Study design: the art of compromise
From our exercise in budgeting this week for the term project, study design appears to be an art of balancing tradeoffs. Molecular reagents and consumables, like tips and PCR plates, add up unbelievably quickly! Add in the number of samples from your “dream” experiment, and the thought of processing 600+ samples is daunting and expensive. However, to thoroughly address a scientific question, one needs a thorough sampling regiment with effective coverage of the study range; as well as proven materials and an adequate number of replicates. In the molecular world, strong preferences for materials and reagents form. Sometimes the least-expensive options are not worth the savings and may even negatively impact the quality of assays and efficacy of results. As an undergraduate, I knew that these materials were quite costly, but never put together my own budget. For the eDNA term project, my group elected to remove an entire treatment series due to cost. While it would be nice to have 72 samples spanning 4 treatments rather than 54 samples spanning 3 treatments, is there any reason to take more samples than you can afford to process? It came down to an extra extraction kit vs an extra aliquot of Taq polymerase (an essential and pricy component of our qPCR assay)- and we would not have enough Taq to process the samples. These are valuable lessons for early-career scientists to learn. How do you know if a compromise you’ve made, especially one for budget or time, will still produce acceptable results beforehand? Another question that has been floating around the lab is which DNA extraction methods to use for what sample types? We (and other labs) have done some testing on this and it seems that certain protocols work better for difficult-to-lyse cell types, like diatoms, then animal tissues. Would it be appropriate to analyze an eDNA sample for kelp or seal presence when a sample’s extraction was optimized for, say, diatoms? Or vice-vesra? I suppose more testing is in order to see if extraction type biases eDNA samples (metabarcoding and single-species).