Due to the continuously increasing amount of data, it is also important to analyse the data in a time-saving, efficient, and effective manner and to ensure data traceability. For this reason, JMP was chosen as the preferred statistical analysis software and a JSL-based script was implemented to analyse the data during the technical performance studies (TPV) in assay development. TPV studies are submission relevant performance studies to claim product performance for several performance indicators. Up to 30 TPV studies are required during product development to proof product performance.
A JMP data management and analysis script containing several built-in scripts was written and validated in JMP to visualise and calculate the data with corresponding JMP platforms and to generate outputs for submission-relevant documentation.
The presentation shows how the script automates processes, thus helping the technicians with assay development during their daily work. After listing the advantages and disadvantages with this approach, it demonstrates how to attempt to compensate for the disadvantages, using a video clip to show how the script works.
We would like to show you with this poster how we have saved time in assay development through our data analysis, improved data quality, and data integrity in order to bring our products to the market faster. My name is Anika Cobernuss-Rahn, and I work for Roche Diagnostics in assay development as product care lead for blood screening products.
My name is Britta Silbersack, and I worked at Roche Diagnostics, too. I'm a senior scientist and lead the JMP support team in assay development. I want to start our poster presentation with the following sentence of Seneca. "It is not that we have so little time, but we lose so much. Life is long if you know how to use it."
Time is a very valuable commodity in today's world. We, in assay development, have been looking for a way to analyze our constantly growing amount of data efficiently and effectively. Our journey, which began a few years ago, can be compared with the flight of a space shuttle. A space shuttle needs a lot of energy to get into space. There are also many ways to get into space. Our way in assay development with automated data analysis has been similar as described in these five pictures.
I still remember clearly when I used to analyze my data using Excel spreadsheets. It took me a while to create charts that I was satisfied with and to find correct copy and paste errors or formula errors in a data analysis. A lot of time was spent analyzing and checking the data to avoid and rectify errors.
It wasn't just me struggling with this, but many of my colleagues as well. As a result, we started looking for a solution and found it in automating data analysis to reduce effort and simplify the process. The software we were looking for had to meet certain requirements. We wanted to be able to view our data similar to Excel spreadsheets, perform statistical analysis easily, and quickly and effortlessly create and modify the necessary charts. It also seemed beneficial to be able to write small programs. That's why we chose the JMP software, and so our journey begins with the launch of the space shuttle.
As I mentioned earlier, a space shuttle needs a lot of energy to successfully reach the orbit. That means it needs enough fuel which is stored in different tanks. We also need a lot of energy to implement the automation of our data analysis in JMP. This energy comes from our assay development team who agree with us, support us, and must also implement the automation.
We have highlighted the advantages that automation offers. We have also shown how the project timeline is affected when our data analysis is automated. Errors are avoided, and the time required for the analysis is significantly reduced. Automation increased also the data integrity and data quality. After introducing the software and presenting some examples, the assay development team decided to use the JMP software for data analysis. Through this decision, our space shuttle can now successfully start its journey into space.
A software or script is only as good as its usage. How can we convince colleagues to change the long-standing process of data analysis? We try to generate interest among colleagues for JMP by demonstrating how quickly calculations and chart creation can be done. We offer training courses to teach them frequently, performed analysis, and answer their questions. Furthermore, we established a JMP power usage team consisting of individuals with more experience in using JMP. Their goal was to assist users in resolving any issue that raised during their data analysis.
The last energy booster for our space shuttle concerned our JMP power user team, which was integrated into other projects in addition to the JMP support. They didn't always have time to carry out the JMP training courses alongside their daily work. We have therefore set up a JMP power user website with the following content.
First, introduction of all power users to distribute the support workload across all power users. Second, the training material of the training courses has been public, so the user can get the training when it fits for them without a power user. Third, creation of cheat sheets for the evaluation of specific analysis topics, for instance, sensitivity or linearity. This allows the user to conduct the analysis themselves.
Fourth, we also create short and simple training films for special JMP functions such as concatenate of a data table, recode or text to column, etc. Last but not least, we want to bring people together in a JMP scripting club who are interested in learning the JSL language. This allowed people to share their knowledge and ask for help when writing their own scripts. Additionally, scripting requests from other users are collected in a wishlist to give the JMP scripters additional opportunities to improve their scoping skills.
Our space shuttle has now successfully reached the orbit.
However, in order for it to continue flying successfully in space, the automation of data needs to expand. This means that existing scripts must be maintained and new ones written. We also wanted to use scripts for our technical performance studies, abbreviated as TPV studies, which are required for claiming product performance for various performance indicators. For this, the script needs to be validated.
To handle these tasks, we have formed a JMP scripting team that maintains existing scripts and modifies them when necessary. They also implement new projects into the existing script and develop new scripts that are available for other data analysis. Here on the poster, the journey of automating our data analysis comes to an end. The use of our scripts has now become part of our data analysis process. This has standardized and simplified our analysis reports while reducing the susceptibility to errors.
However, the journey is not yet complete. Various changes still await us, as mentioned here on the poster, decreasing interest in data, leading to a quiet reliance on support and decline in knowledge and interest in statistics. An example script is presented in the following video.
Our script can be divided into three parts. The first part is a user import. It consists of two user input boxes that prompt for detailed information about the assay's study and location of the data files. Without this information, it is not possible to use the script.
The second part is the creation of the data table. All files were concatenated to a data table and initial calculations were performed. Built-in scripts that can be executed by users have been saved on the top left-hand corner. These built-in scripts vary from study to study and can be executed once the data is concatenated. If the users need additional calculations, they can also use the data table which contains all of their data.
The third and final part is the output of the script. The reports generated by the script are saved in the output folder and are available to the user for documentation purposes.
With these steps that we have presented in this poster, we were able to reduce the development time of an assay and increase data integrity and data quality. As a result, we can bring our assays to the market faster. Thank you for your attention and your interest in this topic. We wish you every success if you want to integrate the automation of your data analysis using the JSL language.
