Last week, I wrote about datasets getting lost when a biotech team switches from highly flexible, highly exploratory work to more constrained and repeatable work. This is mostly a technical problem - if you do the technical work of leaving enough breadcrumbs to find the data later, you should be OK. But there is another, more cultural, shift that happens around the same time. Or at least it aught to happen, whether or not it usually does. A few weeks ago I wrote about how your experiments need names. This week, in my ongoing series about issues that biotech teams run into at specific points in their evolution, I want to explore why it’s so hard to make those names consistent.
If you ask 10 bench scientists what’s the best way to name things like assays, protocols or experiments, you’ll get 11 answers. And while this is fine during that early exploratory phase where you’re going to throw away most of the data, it becomes a problem when you start wanting to keep data around. If half your experiment names start with the date when planning happened and the other half have the date when the lab work started, but at the end of the name… you’re going to have a bad time.
Ultimately, every naming convention is going to have benefits and drawbacks. What your bench scientists choose to go with will come down to personal taste and the subtleties of how each scientist likes to search for things. And technical limitations usually play a role as well: It’s generally considered a bad idea to include metadata like dates in filenames. But if you don’t have an alternative way to look them up by the metadata, the filename may be the only place to put it. And even if you do have another way to look it up, bench scientists who are used to stuffing metadata into filenames will have a hard time dropping that habit.
But let’s put aside, for a moment, the goal of agreeing on *good* naming conventions, and just talk about agreeing on *any* naming convention. And let’s assume you’re at that point where you’re switching from highly exploratory to more repeatable experiments. During that exploratory phase, it didn’t make sense to try and define consistent naming conventions because nothing else was consistent. So up until now, every bench scientist has just been naming things however they want.
So here you are with your 10 bench scientists who have been following 11 different naming conventions (if you’re lucky). None of the 11 naming conventions are objectively better than any of the others. But you need to somehow convince the 10 bench scientists to all switch to one of them.
You’ll quickly realize that the problem isn’t convincing the scientists that you need consistent conventions. They’re smart. There’s a clear benefit to consistency. They get it. In fact, the problem isn’t even convincing them to change. Sure, maybe sometimes one or two will be stubborn and refuse. But these are mostly just personal preferences, not deeply held beliefs. In my experience, this is rarely the problem.
Instead, most often the problem is that no one feels empowered to make the decision. Individual bench scientists don’t want to impose their preferences on other teams that they have nothing to do with. The CSO or Head of R&D who theoretically has this authority, doesn’t want to micromanage. And the data team doesn’t want to meddle in the lab’s business. Maybe if you have a lab informatics team, they might be willing to make a decision, but often they only feel empowered to enforce these kinds of decisions, not impose them.
So it turns out that before you can make this operational decision - what naming convention to use - you first need to make a meta-decision: Who has the authority to make the operational decision?
In theory, the meta-decision should be easier to make than the operational decision. Since everyone can agree that *any* operational decision will be better than no decision, it shouldn’t matter who makes the decision. Of course, any meta-decision that dips into authority and politics can be tricky, but at least it’s something that won’t feel like micro-managing to a CSO or Head of R&D who clearly has the authority to delegate.
The hard part is usually recognizing that this meta-decision is the bottleneck. But that gets significantly easier once you know what to look for. And after reading this, maybe you will.
So, now the really important question: What meta-decisions are getting in your way today?
Thanks for reading this week’s Scaling Biotech! I really appreciate your continued support, and I read every comment and reply.
As a reminder, I offer several services to help connect biotech teams with tools, practices and expertise to make their organizations more data driven.
For Biotech Startups: My Stack Audit will help you create a shared understanding of expectations and gaps around data and metadata.
For Software Startups: Go-to-market consulting to help you find your ideal customers and communicate how you’ll solve their biggest problems.
For IT/Informatics Consultants: White papers and case studies to define and demonstrate your specialized expertise.
For more information about any of these, send me an email at jesse@merelogic.net
Great article! The biggest challenge with naming conventions that we encountered was not recognizing that we needed to standardize our naming conventions or leadership supporting the implementation, but the actual follow through by all team members generating the datasets.