Biotech data is a three-dimensional problem
While building out the framework for the System Evaluation program I’ve been plugging the last few weeks, I’ve started thinking about data infrastructure in terms of three angles, each of which gives you a different view of what you have and what you need to build. As I talk to folks at different biotechs about what they’re looking for and what they think a design should look like, I’m realizing that they’re often looking at it from just one of these angles. But to get a complete view, and to communicate a complete design, you need all three. So this week, I want to describe what they are, how they’re different, and how they interact.
The Stack View
Those of us who come from the Tech world are used to thinking of infrastructure as a stack of components. And by a stack, I mean literally layers stacked on top of each other - Often it’s a storage layer at the bottom with databases and file storage, then a backend server/API layer above it that controls access to the storage layer, and a client/app layer on top that users interact with. This is a good way of organizing the technical components and understanding how they communicate with each other. In other words, it tells you what you need to build and maintain.
However, this view turns out to be pretty bad for understanding the different kinds of needs that a biotech faces. From this angle, each interaction looks like a sequence of bounces between the layers. And the bounces are roughly the same, whether you’re validating a target, calling hits or anything else. These different activities often use the same components of the stack but in different ways that you can’t see from this angle.
The Pipeline View
If you start to map out these different kinds of activities that a biotech startup needs to support, you end up with more or less the traditional view of the pipeline for your corner of biotech. For biopharma/drug dev that’s something like target identification, target validation, hit identification, hit to lead, optimization and so on. This is the complete opposite of the stack perspective, in that it doesn’t show any of the technical infrastructure, only the use cases.
In fact, you might even argue that this shouldn’t be considered a system diagram at all, since it doesn’t show us anything about the system. And sure, just writing out the steps isn’t a design. But it is important to factor into the design how the system will support each of these activities, and how it looks different (or the same) for each of them. However you end up drawing this angle, you need to capture these differences in your overall design.
The Decision Making View
The Stack View and the Pipeline View are at opposite ends of a spectrum from only showing what’s common to only showing what’s different. They also live at completely different time scales: A path through the stack diagram can take milliseconds. A path through the pipeline diagram takes years. These differences make them nicely complementary, but they also make it hard to connect them. So I’ve started working with an angle that I think sits in the middle of the scale and complements both.
To understand this angle, I’ve found it helpful to frame every step of the pipeline in terms of decision making: Deciding what targets to evaluate, what to screen, what to follow up on, and so on through deciding what goes in the IND (or whatever the equivalent is for your area of biotech). You can break each decision down into a common set of steps - for me, it’s planning, experiment, analysis, decision - and then further break them down into sub-steps and sub-sub-steps. This is the angle I used for my post about the process of analysis a few weeks ago.
The time scale of this diagram is in between the stack and the pipeline - A path through it takes weeks or months. Each sub-step takes hours to days. Like the pipeline diagram, this splits up the stack diagram - different sub-steps will use the same parts of the stack in ways that are harder to see from this angle. In the other direction, it squashes some of the information in the pipeline view - target identification, optimization and everything else all follow the same steps. But in my experience, it does less splitting and squashing than the more extreme angles.
In particular, I’ve found that this angle is light enough on the splitting that you can still map pieces to the stack diagram, and it squashes the pipeline gently enough that you can still identify how the needs of the different stages differ.
Conclusion
Ultimately, you need to leverage all three angles to design and build the right infrastructure and processes for your biotech startup. And doing that starts by understanding how they’re different and when to use each one.
To get a better understanding of what your biotech startup looks like from all three angles, sign up for a System Evaluation. I’ll walk you through a detailed rubric and create an insightful report identifying the parts of your data processes and infrastructure you need to address today, and what can wait for later. Sign up today!