The client had a backlog of experimental data across a variety of assay types that was not being captured into the corporate database. Data processing was being performed manually by the scientists in static spreadsheets. A collection of data capture templates was therefore required.

Templates were built in parallel, in order to avoid pauses in progress whilst waiting for feedback. For each template the following generic approach was followed. A meeting took place with the scientist(s) to thoroughly gather requirements and understand the step-by-step experimental procedure with any conditions and results mapped out. On occasion the biologist gave a laboratory demonstration of their experiment. A high-level understanding of the science would be obtained, since this is the key to implementing a successful data capture workflow.

Familiarisation with the client's results and conditions dictionaries and business rules, so as to avoid introducing replicate entries. Checking that a new template is really required or if a current one can be extended. The documented requirements would then be reflected back to the customer with any required follow-up meetings before proceeding with implementation.

There was communication with the biologist throughout the process with demonstrations of the template during implementation, so any shift in requirements could be quickly addressed. Clear interfaces were designed with step-by-step instructions. This is especially important for future scenarios where the scientist who is performing the experiment may not have been involved in the build.

Finally before providing the template to the scientist it would be rigorously tested with both data provided by the scientist and mocked up test files. This case study closely resembles work that Amethyst has carried out for a multitude of clients (small start-ups, large Pharma, Universities and Contract Research organisations). We offer support across a variety of vendor products, including IDBS E-Workbook, IDBS ActivityBase and Dotmatics Studies. We have also written custom code to process data; for example VBA macros embedded into Excel or web applications (using JSP, SQL and JavaScript).

The backlog was cleared with experimental data now being loaded into the database. This removed a huge bottleneck and has reduced the time Biologists spend performing manual error-prone steps. Data across experiments will now be more readily available to any downstream data analysis. This enables Drug Discovery projects to fully utilise the knowledge contained within its experimental results.