Use case #3 Organising Laboratory Data Workflows

• Optimal collection, processing and storage of raw laboratory data
• No more copy/pasting of datasets
• Reproducible, reliable, harmonised data organisation, processing, visualisation and reporting workflows

The context

The typical modern laboratory is crammed with electronic equipment of various ages and complexity, all using different software and reporting formats. Data generated by this disparate equipment must be saved in a secure and traceable manner for future access and use. Ideally, this means automation to avoid laborious, error-prone copy/pasting.

The challenge

Develop an automated process to upload output raw files from lab equipment into a cloud-based system. Make this system platform-independent so that data is easily transferable. Enable the user to examine datasets online or access them easily via API linking. Support the annotation of uploaded datasets so that even researchers not directly involved in data generation can use the datasets.

Our use case: the h-CLAT assay

To see how EdelweissData™ meets this challenge, we took one of the most important in vitro tests for skin sensitisation, human Cell Line Activation Test (h-CLAT)* as our case study.

Datasets are generated in h-CLAT: In both main stages the cytometer is used to read assay plates. The generated raw data is usually organised in tabular form (CSV). These raw datasets need to be further processed to calculate the CV75, EC150 and EC200. Additionally, one may wish to store datasets with supporting information generated at different stages of the assay (e.g., sample information, reagent and instrument information, solution preparation).

Disadvantages of the generated datasets: CSV or Excel files are poorly structured (e.g., information is stored partly in rows, partly in columns) and stored in separate files on a local disc. This makes it difficult to search within or between such datasets, as well as difficult to automate calculations based on this kind of dataset.

How could this be fixed?
The machine generated datasets should first be annotated. For example, assay plate configuration should be manually prepared by the user and merged with the readouts dataset. In this way, a computer can “understand” what is the content of each assay well, and dataset processing can be automated.

By a clear separation of data (measurements within the assay) and metadata (information about the assay as a whole), and structuring the information within them, searching within and between the datasets can be largely improved.

Where EdelweissData™ comes in
Using the h-CLAT assay, we demonstrated that EdelweissData™ allows for the secure storage of annotated tabular datasets and easy access of the datasets via API link.

To demonstrate this workflow, we prepared a Google Colaboratory notebook to guide the user through the various steps of data storage and analysis. (In the actual implementation, the Google Colaboratory notebook, which serves only for quickly developing prototype solutions, would be replaced by a proper user interface.)

The final results can be additionally annotated with the sample information and stored in the EdelweissData™ or could serve as an input to the SaferSkin™ application, which calculates skin sensitisation potency based on the three defined approaches. Note that other inputs for the tested compound (results of the DPRA and KeratinoSens™ assays) could also be efficiently found with the search among datasets supported by the EdelweissData™.

The aim of this workflow and related use cases is to help automate the handling of data generated by labs (research, CROs, etc.) within in vitro studies. The workflow is supported by EdelweissData™ and Google Colaboratory notebook tools and addresses the data file processing, storage, analysis and reporting. The data consist of experimental results enriched with metadata (details on how the data was produced). This approach follows FAIR (Findable, Accessible, Interoperable, and Reusable) data principles and aims to enhance the accessibility and usability of the data, making it accessible in real-time, well-annotated and reliable.

---