In this project we developed a screening data capture web application for invitro biologists. Off-the-shelf software for data capture was not suitable in this case for reasons of existing infrastructure constraints, specific data processing requirements, and the need for flexibility in assay definitions.
Existing infrastructure constraints.
There already was a screening database in place, with data. A commercial screening data capture solution would have required a substantial customization and integration effort to be made compatible with this database and query applications. Moreover, the database was to be replaced in a few months by another custom data repository tightly integrated with a suite of previously built in-house data query applications. The Client instead decided to develop a light data capture web application and connect it to the existing database.
Statistically sound fit algorithms for noisy data.
Another critical requirement was to perform dose-response curve fitting with a specific multi-step algorithm because of the nature of the data in these assays. The curves were unsuitable for typical automated fit tools: either there was no fit in too many cases and time-consuming experiments had to be redone again and again, or artificial curve constraints were required to get any fit. The sheer amount of screening data coming in each week made manual fitting impossible. However there was a published algorithm known to work well with these types of assays. It was (and still is) unavailable in commercial screening data management software whereas it runs very well as part of our Client's screening data capture web application.
Flexible assay definitions.
The biologists developed new assays on a weekly basis, so new capture templates had to be designed accordingly. A significant number of new assays would require unique templates and data processing steps, which is not easily done with Excel templating approaches. In many cases input data contained errors or omissions: automated and intelligent user feedback was required in the data capture process. An onsite template / web application developer rather than an Excel expert was needed.
Data capture, review, final approval.
The Client teams put in place a data review and approval process to ensure that all results could be compared regardless of experiment capture date, person or method. The capture needed to be transparent, consistent, and reproducible. Reviewers wanted to have an internal secure web portal where requests would be queued; they also expected email notifications as new assay results were being submitted for review/approval. We were asked to implement this exact review/approval process in the web application.
Research IT closely collaborated with invitro biologists.
Research IT discussed the data capture needs with biologists. Together they made the decision to try an in-house solution as a better fit to their requirements than buying into yet another large screening informatics system. Since all the biologists are co-located with the research IT team, the collaboration was largely informal and ideas could be validated very quickly.
