Scope of a typical dissertation project in a clinical PhD program

Lab members should strive to thoroughly read and process at least 1 project-related paper per week. If the trajectory laid out above is followed, ~14 months will pass between the time a student informallymed joins the lab and proposes a dissertation project.

As mentioned earlier, this time should be spent learning the basic lab procedures, techniques, and models, and thinking about how these techniques and models might be applied to a new project. Importantly, lab members should start collecting and recording data in very small experiments / replications (e.g., gender differences, longevity after insults, dose response curves, assay parameters, food consumption, etc.), organizing the collected data in a spreadsheet, and making graphs. These should be included in your weekly lab notebook printout for review in our meetings. By the time you are ready to propose, data collection and analysis should be second nature and all the kinks should be worked out. Note that although the following section discusses Drosophila projects, similar principles apply to rodent or human studies (but the sample sizes will be significantly smaller, so these designs cannot include as many factors).

Assuming you have read about 1 relevant paper per week, your proposal should have roughly 50-60 references. It should also contain enough pilot data / graphs to demonstrate that you can collect data for each of the dependent variables that you propose to use. If your experiment is based on previous experiments, you should show (with a minimal set of data) that you can replicate those results. Note that you should continue to read more project-related papers at a pace of (at least) about 1 per week.

Again, if the trajectory laid out above is followed, ~19 months will pass between the time a lab member proposes a dissertation project and defends. At least 16 of these months should be actively spent collecting data, allowing for ~3 months to analyze, interpret, and write up the results. Of course, if you opt to push the dissertation defense to the official deadline, that’s 31 months.

Therefore, a completed dissertation project in a clinical department should reflect at least 16 months of steady data collection (since the proposal) and at least 33 months of literature consumption / synthesis (since joining the lab). Generally, this should end up in the ballpark of ~1000-2000 flies in one large balanced design or multiple, smaller, stepwise projects, and at least 140 referenced papers. The actual data and research throughput will usually ebb and flow over this period, but this averages out to testing ~60-70 flies / month (~15 flies / week). By the time you are done, you should literally be an expert on the very specific topic that you have chosen.

The following example designs would provide the backbone of a project with sufficient scope for a PhD dissertation in this lab (i.e., publishable):

Design 1:

• Gender: 2 levels (male vs. female)
• Age: 2 levels (young vs. aged)
• Diet: 2 levels (control vs. polyphenol)
• Insult: 3 levels (control vs. high/low irradiation, anesthesia, sleep deprivation, etc.)
• Mechanism: 2 levels (control vs. agent to block putative polyphenol effects)

Design 2:

• Gender: 2 levels (male vs. female)
• Diet: 6 levels (polyphenol: 0%, 5%, 10%, 20%, 40%, 80%)
• Insult: 2 levels (control vs. irradiation, anesthesia, TBI, etc.)
• Mechanism: 2 levels (control vs. agent to block putative polyphenol effects)

Design 3 (females only)

• Diet: 6 levels (polyphenol: 0%, 5%, 10%, 20%, 40%, 80%)
• Insult: 4 levels (irradiation: 0 Gy, 1 Gy, 5 Gy, 25 Gy)
• Mechanism: 2 levels (control vs. agent to block putative polyphenol effects)

Each of these projects would require 48 groups for a balanced ANOVA design. At least some of these factors will probably be devoted to replicating previous data. With a generous sample size of 20 flies per group, that’s 960 flies overall. Adding another 3-level factor (e.g., Diet Timing: throughout the experiment, before insults only, or after insults only) would require 144 groups - using 14 flies per group would require 2016 flies overall. Additionally, behavioral tests that assess multiple behavioral domains (e.g., motor, memory, recovery time, exploration, etc.) should ideally be performed on the same flies over multiple time periods (i.e., repeated-measures designs). Such a design would require multiple (identical) small replicates to build up to these numbers over the 16-28 month data collection period.