Harnessing QbD, Programming Languages, and Automation for Reproducible Biology

Abstract

Building robust manufacturing processes from biological components is a task that is highly complex and requires sophisticated tools to describe processes, inputs, and measurements and administrate management of knowledge, data, and materials. We argue that for bioengineering to fully access biological potential, it will require application of statistically designed experiments to derive detailed empirical models of underlying systems. This requires execution of large-scale structured experimentation for which laboratory automation is necessary. This requires development of expressive, high-level languages that allow reusability of protocols, characterization of their reliability, and a change in focus from implementation details to functional properties. We review recent developments in these areas and identify what we believe is an exciting trend that promises to revolutionize biotechnology. Biology, Context Sensitivity, and ReproducibilityThe 21st century will be the century of biotechnology [1,2]the economic contribution of biotech is set to grow substantially [3,4] and it has been suggested that biotechnology offers solutions to a variety of crises faced by humanity such as climate change [5], the supply of food [6], energy via biofuels [7], and tackling escalating healthcare costs.However, particularly with regard to the cost of healthcare, a number of authors have argued that the pharma sector is itself in crisis [8][9][10][11]. They warn of a looming danger of significantly diminishing returns on R&D expenditure in the pharmaceutical industry [12], a phenomenon somewhat facetiously known as 'Eroom's law'. This 'law' is purportedly the inverse of Moore's law (which describes an exponential increase in integrated circuit transistor density over time) and denotes an exponential decrease in ROI (return on investment) for investment in pharmaceutical R&D. In an analysis of the underlying causes, Cooke [11] suggested that, in addition to the economic factors, there are substantial problems with the fundamental framework of biological research.Specifically, following comments made by Lord Winston [13] and echoing the rallying cry of systems biology, it is suggested that the basis of pharmaceutical research is too narrow, linear, and gene-centric, and fails to properly account for the true complexity of biological systems.There may be other signs of impending problems: several recent studies have reported low rates of reproducibility across several areas of science including drug discovery [14], psychology [15], synthetic biology [16], medicine [17,18], and cancer research [19]. One recent study estimated that the annual economic cost of irreproducible research in the life sciences is $28 billion [20]. TrendsBiological complexity is a barrier to fulfilling the potential of biotechnology.Large numbers of complex experiments are required to overcome this barrier.Performing such complex experiments requires sophisticated software and hardware.New programming languages and software tools for this are developing quick...

Linking Engineered Cells to Their Digital Twins: a Version Control System for Strain Engineering

“…Electronic Lab Notebooks and other online applications that use the cloud to store and publish the data and experiments generated are a partial step in this direction although they are generic for lab operations rather than specific to the genetic programming of organisms. Thus new tools and software tools have been called for (Sadowski et al, 2016).…”

Sequential Multivariate Cell Culture Modeling at Multiple Scales Supports Systematic Shaping of a Monoclonal Antibody Toward a Quality Target

“…Many of these trends are already well‐established in the “older” industries such as automotive. Therefore, in order to eventually fulfill the standards and goals of the industry 4.0 era, the methodologies and tools associated to previous trends must be further developed and extensively utilized in the biopharmaceutical process industry …”

“…Therefore, in order to eventually fulfill the standards and goals of the industry 4.0 era, [29,30] the methodologies and tools associated to previous trends must be further developed and extensively utilized in the biopharmaceutical process industry. [31][32][33][34][35] In the last decade, titers in mammalian cell cultures could be increased to values as high as 5 g L À1 in production, so that the identification and modulation of the critical quality attributes (CQAs) of the product became increasingly important. [1,5,36,37] The quality fingerprint of therapeutic proteins is quite complex and includes versatile biological patterns such as the glycosylation and charge variant profiles as well as the aggregated and low molecular weight forms, all of which are highly important for the efficacy, potency and safety of the drug.…”

Intensified design of experiments for upstream bioreactors

“…In light of the Process Analytical Technology initiative and the promoted Quality by Design paradigm it is of critical importance to show that the impact of all the process parameters (factors) on the process are understood . High‐throughput platforms and single‐use equipment have found increasing application in recent years allowing parallel studies of entire DoEs , which has the potential to reduce process development / optimization timelines significantly. The data that results from DoE are typically investigated using multivariate data analysis methods, in particular Response Surface Models are popular .…”

“…High‐throughput platforms and single‐use equipment have found increasing application in recent years allowing parallel studies of entire DoEs , which has the potential to reduce process development / optimization timelines significantly. The data that results from DoE are typically investigated using multivariate data analysis methods, in particular Response Surface Models are popular . These approaches work well in the vicinity of the process optimum since the solution surface can be approximated by quadratic functions, but the time‐course of every experiment is typically reduced to a static representation.…”