A cell culture online data analysis tool combining data from multiple experiments and databases

Fed-batch cultures of mammalian cells are often used to synthesize biopharmaceutical proteins for therapeutic use. Small scale cell culture experiments are often conducted in order to develop and optimize process variables that affect the cell growth and product quality attributes of the proteins produced. Typical data collected during these experiments are time varying online process data such as temperature, pH, cell density, and batch-associated data such as the product quality attributes for proteins produced from the specific batches. Data from these experiments had been typically stored in multiple databases for the purposes of data integrity, which required that they be downloaded to local computers for researchers to combine and compare this data across experiments. In particular, databases that hold high-temporal-resolution process data are only superficially linked to experiment registries where batch-associated variables are stored. By establishing virtual database infrastructure via JBoss EDS (Enterprise Data Services) and using JDBC (Java Database Connection) protocol for connectivity, we created a MATLAB application that allows the user to select experiments from the registry and then view process variables associated with those experiments. The primary method of experimental analysis is often done by comparing a variable's time profile both quantitatively and qualitatively; thus it is critical for these trends to be easy to compare, even if the trends originate from different experiments. As a result of these efforts, users can now visualize a variety of variables' time profiles from any registered experiment and dynamically group these trends to discover possible effects of experimental condition.