This blog post discusses Step 1 of the Descign’s Master Plan as mentioned in our previous post. We explore WHY research documentation is a big issue and the current solutions addressing this problem. We also propose Descign’s approach which is to automate research documentation.
A quick recap of our 4-step master plan to exponentially accelerate research is:
Step 1 - Automate Documentation
Step 2 - Enable Seamless Collaboration
Step 3 - AI co-pilot
Step 4 - Solve Moonshots
Let’s dive right into it.
Research is expensive and depending on the domain it can easily cost millions of dollars. This is because it requires highly skilled people (scientists, technicians, managers), expensive infrastructure (labs, instruments, storage), high-end consumables (lab ware, buffers, chemicals, biological) and large amounts of data storage and compute. We spend over $2 Trillion on research globally, yet due to poor documentation we have a major reproducibility crisis - Only 1 in 10 papers can be reproduced.
The reproducibility crisis can be avoided if we carefully document and share the entire lifecycle of an experiment. The lifecycle of an experiment is a collection of all the data and metadata generated in each iteration of an experiment, including information on why, how, when and by-whom an experimental iteration was performed. Documenting the entire lifecycle of an experiment is critical because we rarely get the desired result in the first attempt. It’s normal to perform an experiment multiple times (10-100s of times) varying and adjusting parameters until we reach a satisfactory conclusion. Unfortunately, only the successful experiments are reported in publications for convenience and space limitations. And although a publication format that encourages the dissemination of experiment life cycle would benefit us greatly, scientific publications have barely evolved in the last 350 years. The problems with the current scientific publication format could be a separate blog series. Anyways, in summary, what we end up reading in a publication is only the tip of the iceberg i.e. the final successful experiment. And it is almost impossible to reproduce an experiment with only that information. This leads us to a reproducibility crisis. Reporting the entire life cycle of an experiment will help the research community reduce redundancies and hopefully discover insights which are now locked away in a lab notebook or a pendrive.
Current research tools like Electronic Lab Notebooks (ELNs), Lab Information Management Systems (LIMS), Lab Inventory Management Systems (LIMS), Lab Information Systems (LIS) Scientific Data Management Systems (SDMS), Lab Execution Systems (LES) etc. do not follow any good documentation format. They either offer their own proprietary format or a “flexible” format - both of which come with their own set of problems. Furthermore, no single tool meets all the requirements of a researcher. A lab needs to buy, learn and maintain multiple non-interoperable solutions. This results in data being locked up in different tools. As a consequence, researchers need to spend a lot of time on manual data entry to maintain up-to-date systems. If there is no incentive in place, documentation takes a hit.
Short Answer - Automate documentation.
Long Answer - Our team at Descign used a classic business framework called ‘Jobs to be done’ (JTBD) to solve the documentation problem. Applying the JTBD framework to a researcher's activities allowed us to understand the research workflow from a first principles perspective. The framework “revealed'' that a general research workflow consists of four stages: Search, Design, Execute and Analyze. In the Search Stage, a researcher reads and comes up with an idea or a hypothesis. In the Design stage, they specify experiments to test the hypothesis. In the Execute stage, they plan and perform the experiments and in the Analysis stage they assess the results. Even complex R&D workflows for different domains like medicine, biology and chemistry can be grouped into these four stages. Researchers working in synthetic biology already use something very similar called the DBTL (design, build, test, learn) cycle.
We used our learnings from the JBTD exercise and developed a platform which allows digitization of the entire research workflow. It captures all the stages (search, design, execute and analyze) on the platform and assists the researchers in managing their activities in the different stages of their workflow. As the experiment progresses through these stages, the platform automatically documents the entire lifecycle in a structured manner. Structure and detailed documentation enables researchers to understand each other's work and enhances collaboration. Research is a game best enjoyed in multiplayer mode.
Simply put - Our approach is to shift the burden of documentation from researcher to software.
We are confident this will
In the next blog, we will discuss Step 2 - Enable Seamless Collaboration. If you like to learn more about our approach, please drop us a message. We would love to discuss it and improve it. Developing a platform which can accelerate our understanding of the natural world is best enjoyed in a multiplayer mode too. So connect with us on LinkedIn, Twitter or drop us an email at firstname.lastname@example.org.
Want the lastest from our blog straight to your inbox? Subscribe to our newsletter by providing your email.