R&D leaders face growing pressure to innovate quickly and cost-effectively. However, manual paper-based data collection methods stifle innovation, while outdated systems slow productivity and introduce data quality and traceability issues. We explored this topic in our webinar: Revolutionizing Research: Enhancing Experiment Efficiency with an ELN.
Read on for a summarized preview excerpt of the webinar or skip it and watch the full video.
In this webinar we explore the history of scientific documentation, the importance of data quality, and the benefits of using modern ELN technology to advance innovation.
The Webinar Agenda
- The Modern R&D Challenge: In Introduction to the Current Innovation Landscape
- Dive Deep into Data Quality and How It Can Increase R&D Output
- Q&A with Jake Schofield and Barney Walker
- Practical Strategies: Gain Actionable Takeaways and Practical Strategies for Streamlining Science and Experiment Data Capture
Summarized Transcript Excerpt
Jake Schofield: Starting with the challenges of R&D, at the core of the issue we’re discussing today is the fact that R&D is slow and expensive. Research leaders agree it’s critical to reduce product development cycles. This isn’t just about cost savings; it’s also about getting to market faster and increasing competitiveness, driving more profit. How do we address these challenges?
When we think about optimizing the efficiency of an R&D organization, there are three main areas to focus on. Firstly, individual productivity: making scientists’ daily tasks quicker and easier. Secondly, interpersonal efficiencies: removing friction and unnecessary steps in collaboration.
The third piece, specific to R&D, is what we’re calling experimental efficiency. This refers to how efficiently you can achieve results with the fewest experiments or steps. Physical experiments in the lab are costly in terms of time and resources. If you can reduce the number of experiments needed to achieve the same results, the savings can be significant.
R&D is not a linear process like manufacturing; it’s a journey into the unknown. There will always be uncertainties and unexpected discoveries along the way. The key question is: what can we do to reduce the unnecessary diversions and loops in the process to streamline it as much as possible?
There are three primary sources of R&D inefficiency:
- Intrinsic Uncertainty in the scientific process, leading to navigational wrong turns.
- Human Error, resulting in the need to repeat experiments.
- Lack of Reproducibility, which is essential in commercial settings. There’s ample research showing that the lack of reproducibility is a significant problem, costing the scientific industry somewhere around $20 billion. Reproducibility is crucial for validating and confirming experimental results. It is what makes science, science.
How does an ELN fit into all of this?
Barney Walker: In response to that question, I believe ELN plays a significant role. It holds the potential to drive efficiency gains across all the various aspects we’ve discussed. ELN not only enhances individual productivity and fosters effective collaboration, but critically can also improve the experiment efficiency of the overall R&D process.
Modern ELNs can improve experiment efficiency by tackling all three of the sources of inefficiency you raised: from human error, the reproducibility issue, and even some of the intrinsic uncertainty of R&D.
The key to these efficiency improvements is data quality.
When we refer to data in scientific contexts, it entails not only the raw measurements, like plate reader measurements, NMR Spectra, and microscope images, but also all the associated metadata about the experimental procedures and conditions used to generate that data.
When it comes to this kind of data, the 3 main determinants of Data Quality that we are concerned with are Accuracy, Completeness and Structure.
If experimental details are not accurate in the experimental record, then when someone tries to repeat the experiment they are bound to fail to reproduce the results successfully..
Similarly, if the record is not complete crucial details may be missing which make it harder to reproduce a result.. This should includes who conducted the experiment, what procedures were followed, and even, when and where they were conducted, as these factors can influence outcomes significantly.
It should also include detailed information about the materials used, such as precise quantities and specific batch or lot numbers, which can impact subsequent attempts at replication. Understanding the lineage of lab-produced samples is vital for tracing potential contaminations or rectifying errors in the process. Additionally, recording the usage history of materials, considering factors like freeze-thaw cycles, is crucial, particularly in molecular biology experiments.
It involves noting the devices and instruments utilized, including the model, specific settings, calibration, and servicing history, as variations in these parameters can affect outcomes.
Both controlled variables like reaction temperatures and durations, and uncontrolled factors such as lab temperature and humidity, must be documented as well.
Lastly, attention must be paid to the analytical methods employed, as the interpretation of raw data often relies on complex analyses and statistical tools. Failure to document these methods accurately could lead to inconsistencies in results, wasting valuable time on unnecessary experimentation. Collecting such comprehensive data not only facilitates reproducibility but also streamlines troubleshooting efforts in case of discrepancies, ultimately enhancing the efficiency of the entire R&D process.
Now, let’s delve into the final aspect of data quality: its structure.
What exactly do I mean by structure, and why is it so valuable? Well, let me illustrate with an example. Imagine we have meticulously recorded every detail of our experiments in a series of paper lab notebooks. While these records may be accurate and comprehensive, they’re essentially trapped within the confines of those notebooks, limiting their utility.
Contrastingly, if we organize our data in a structured format, such as a relational database, we unlock its potential for broader applications. By structuring our data, we can leverage it for statistical modeling, AI, and machine learning, enabling us to make informed predictions about future scientific endeavors.
Consider a scenario from molecular biology, the field I studied. In the molecular biology lab, we often conducted hundreds of PCRs, sometimes in a single day, across various lab members. Crucial details like primer sequences and annealing temperatures would be scattered across different lab notebooks or digital records, lacking cohesion or context.
Now, imagine if we had all this data structured in a central database, detailing every PCR experiment conducted along with its specific conditions. We could then utilize even basic statistical models to predict optimal conditions for future PCR experiments, streamlining our research efforts without the need for complex AI or machine learning algorithms.
It’s all about streamlining the process, cutting down on those extra steps needed to fine-tune reaction conditions, and minimizing manual troubleshooting.
Our own historical lab-generated data, accumulated painstakingly over years, is a potential goldmine that could allow us to make sharper predictions and pursue more efficient experimental paths.
Barney Walker, Labstep Head of Product, STARLIMS
Every lab out there holds a treasure trove of historical method data, a wealth of insights from countless experiments that could revolutionize our approach to future projects. The catch? Much of this invaluable data remains locked away in formats that are frustratingly difficult to access and utilize. It’s high time we break free from these constraints.
This brings us to the topic of electronic lab notebooks (ELNs).
It’s clear that comprehensive documentation is the backbone of the entire R&D process, from initial planning to final analysis. Yet, despite all the technological advancements we’ve seen, lab documentation practices seem to have hit a plateau. As recently as 2017, a staggering 90% of scientists were reporting that they were still jotting things down with pen and paper.
In some ways, our data documentation practices have actually taken a step backward over time. In the 1800s, the lab book was a complete record of every observation and data point associated with an experiment.
Fast forward to the digital age, method details might still find their way onto paper, however the data itself ends up scattered across various digital platforms, losing its coherence and context. Even if someone prints out a graph and sticks it in a book, critical analysis and details could be lurking in spreadsheets or folders on different computers.
The need for ELNs arose partly due to these challenges. Early ELNs were essentially just digital replicas of paper notebooks. While they offered benefits like digital storage, remote access, searchability, and easier collaboration and organization, they didn’t effectively address data quality or improve R&D efficiency.
The problem with these early ELNs was that they added to the documentation workload for scientists. They were often cumbersome and difficult to use, making it harder to record details accurately. This increased the likelihood of incomplete or inaccurate data. Additionally, the delay between conducting experiments and documenting them in the ELN could lead to memory errors or mistakes. The true potential of digital technology in the R&D space has still yet to be fully realized.
Continue the webinar to learn about modern ELNs address these aspects of data quality by providing integrated features that streamline data collection, reduce manual entry burden, and support completeness and accuracy. These features include:
- Integrated protocol execution system optimized for mobile devices, which acts as a data collection form and supports real-time documentation.
- Integrated inventory system for linking materials, samples, and reagents to experiment records.
- Integrated devices module for linking devices used in experiments and capturing device data.
- Reproducible code capsules for embedding analysis code within experiment records.
- Structured data fields and APIs for interoperability and reuse of data.