Key terms and concepts in bioprocess engineering, scale-up, economics, and physics-informed AI modeling. Definitions reflect how these concepts are implemented in the Augur platform.
The concentration of a target product (protein, metabolite, or biomolecule) in the fermentation broth, typically expressed in g/L. Titer is one of the three key bioprocess performance metrics (T.R.Y. — titer, rate, yield). Higher titer generally reduces downstream processing costs per gram of product.
The volumetric productivity of a bioprocess, measuring how much product is produced per unit volume per unit time, typically expressed in g/L/h or g/L/day. Rate determines bioreactor throughput and directly impacts manufacturing capacity and capital efficiency. Augur predicts production-scale rates from lab data using physics-informed models.
The mass of product obtained per mass of substrate consumed (g product/g substrate), reflecting the metabolic efficiency of the organism. Yield is constrained by thermodynamics and metabolic pathway stoichiometry. In Augur, yield is predicted alongside titer and rate as part of T.R.Y. metrics.
Abbreviation for Titer, Rate, and Yield — the three fundamental performance metrics used to evaluate bioprocess productivity. T.R.Y. collectively determines the economics of biomanufacturing. Augur predicts all three metrics at production scale with confidence intervals from lab-scale data.
Cost of Goods Sold per kilogram of product — the total manufacturing cost including raw materials (media, buffers), labor, equipment depreciation, utilities, consumables, and facility overhead. COGS/kg is the primary economic metric for biomanufacturing decisions. Augur integrates upstream and downstream cost modeling to predict COGS/kg with Monte Carlo confidence intervals.
A calculation that determines the minimum selling price at which a bioprocess becomes profitable, accounting for all manufacturing costs (COGS) plus margins. In Augur, break-even analysis is integrated into every prediction, helping teams evaluate commercial viability before committing to production scale.
A systematic methodology for evaluating the economic feasibility of a bioprocess, integrating technical process modeling with financial analysis. TEA combines mass balance, energy balance, and equipment sizing with capital expenditure (CAPEX) and operating expenditure (OPEX) estimates. Augur automates TEA by coupling physics-informed simulation with cost modeling.
The process of transitioning a bioprocess from laboratory scale (typically 1-10L) to production scale (1,000-200,000L). Scale-up is not simply making a bigger vessel — fluid dynamics, heat transfer, mass transfer, and mixing all change non-linearly with scale. Augur uses physics-informed models to predict these changes and their impact on process performance.
Volumetric mass transfer coefficient (kLa, measured in h⁻¹) — the rate at which oxygen transfers from gas bubbles to the liquid phase in a bioreactor. kLa is the most critical scale-up parameter because oxygen supply often becomes limiting at production scale. It depends on agitation, aeration, vessel geometry, and broth properties. Augur models kLa changes during scale-up to predict dissolved oxygen profiles at production scale.
The time required to achieve 95% homogeneity after adding a substance to a bioreactor. Mixing time increases with vessel scale — a 2L bioreactor might mix in seconds while a 20,000L vessel can take minutes. Poor mixing creates concentration gradients that stress cells and reduce productivity. Augur predicts mixing time at production scale as part of its hydrodynamic modeling.
The velocity at the tip of the impeller blade (m/s), calculated as π × impeller diameter × RPM. Tip speed is a key scale-up criterion because excessive tip speed causes shear damage to cells, while insufficient tip speed leads to poor mixing and oxygen transfer. Augur maintains tip speed within organism-specific ranges during scale-up predictions.
The concentration of oxygen dissolved in the fermentation broth, typically expressed as a percentage of air saturation. Maintaining adequate DO is critical for aerobic organisms — below a critical threshold (typically 20-30% saturation), growth and product formation decline. At production scale, oxygen supply often becomes the primary bottleneck.
A class of neural networks that incorporate physical laws — such as conservation of mass, energy balance, and transport equations — directly into the model's loss function and architecture. In bioprocess simulation, PINNs constrain predictions to obey Monod kinetics, product inhibition, and overflow metabolism. This physics grounding enables accurate predictions from as few as 5 lab runs, compared to hundreds required by conventional machine learning. Augur uses an ensemble of 5 PINNs with ODE residual learning.
A training approach where the PINN learns to predict the residual (difference) between actual experimental data and an ODE-based kinetic model, rather than predicting absolute values. This means the PINN only needs to learn the deviations from known physics, making it more data-efficient and stable. In Augur, the ODE baseline provides Monod kinetics predictions, and the PINN corrects for real-world effects that the ODE cannot capture.
A statistical framework for producing prediction intervals with guaranteed coverage probability, regardless of the underlying model or data distribution. In Augur, conformal prediction calibrates the PINN ensemble's confidence intervals using held-out lab data, ensuring that stated uncertainty bounds are statistically valid rather than heuristic. This gives process engineers trustworthy uncertainty quantification for scale-up decisions.
A computational model that mirrors a physical bioreactor and its associated processing steps, combining real-time or historical sensor data with physics-based simulation to predict process outcomes. Unlike simple process models, a digital twin continuously learns from new data and can simulate what-if scenarios. Augur creates predictive digital twins from as few as 5 lab-scale fermentation runs using physics-informed neural networks.
A mathematical model describing microbial growth rate as a function of substrate concentration: μ = μ_max × S / (K_s + S), where μ_max is the maximum specific growth rate and K_s is the half-saturation constant. Monod kinetics forms the foundation of bioprocess simulation and is embedded as a physics constraint in Augur's PINN architecture.
The reduction in microbial growth rate caused by accumulation of the product itself. Modeled as μ × Kp/(Kp + P), where Kp is the inhibition constant and P is product concentration. Product inhibition is organism-specific: CHO cells show strong inhibition (Kp = 8 g/L), while E. coli is more tolerant (Kp = 50 g/L). Augur incorporates product inhibition in both ODE kinetics and PINN physics constraints.
A metabolic phenomenon where organisms produce overflow metabolites (ethanol in yeast, acetate in E. coli) when glucose uptake exceeds a critical rate, even under aerobic conditions. This diverts carbon away from biomass and product formation. Augur models overflow metabolism as a 6th ODE state variable with organism-specific critical glucose uptake rates (q_s_crit).
A model describing how reaction rates (including microbial growth) change with temperature: μ(T) = μ_max × exp(-Ea/R × (1/T - 1/T_opt)), where Ea is activation energy and T_opt is optimal temperature. Each organism has characteristic Ea and T_opt values. Augur uses Arrhenius modeling to predict the impact of temperature shifts during scale-up.
All unit operations performed after fermentation to purify the target product to required specifications. DSP typically accounts for 60-80% of total biomanufacturing cost and includes steps like centrifugation, filtration, chromatography, foam fractionation, and formulation. Augur models 9 DSP unit operations with organism-specific templates and integrated cost calculations.
All operations involved in growing cells and producing the target molecule, including media preparation, inoculum development, and bioreactor cultivation. USP determines the titer, rate, and yield that feed into downstream processing. Augur couples USP simulation with DSP economics for end-to-end COGS prediction.
An affinity chromatography technique used as the capture step in monoclonal antibody (mAb) purification. Protein A binds to the Fc region of IgG antibodies with high specificity, achieving >95% purity in a single step. It is typically the most expensive DSP operation due to resin cost ($8,000-15,000/L). Augur's mAb DSP template includes Protein A as the first chromatography step.
A membrane-based separation process used for concentrating protein solutions (ultrafiltration) and exchanging buffers (diafiltration). UF/DF is commonly the final DSP step before formulation. Augur models UF/DF with concentration factor, membrane area, and flux calculations to predict processing time and cost.
A separation technique that exploits the surface activity of biosurfactants to concentrate product in foam. Air is sparged through the fermentation broth; surface-active molecules (rhamnolipids, sophorolipids, surfactin) adsorb to bubble surfaces and are carried upward into a foam phase. The collapsed foamate contains concentrated product (typically 10-50× enrichment) while cells remain in the liquid phase. Foam fractionation is solvent-free, low-energy, and uniquely suited to biosurfactant recovery. Augur is the only bioprocess simulation platform that models foam fractionation as a DSP operation with enrichment ratio, recovery, and cost calculations.
A surface-active compound produced by microorganisms, capable of reducing surface tension and emulsifying hydrophobic substances. Major classes include glycolipids (rhamnolipids, sophorolipids, mannosylerythritol lipids), lipopeptides (surfactin), and polymeric biosurfactants. Biosurfactants are biodegradable and less toxic than petrochemical surfactants, but must achieve COGS below $1-5/kg to be cost-competitive. Applications span cleaning products, cosmetics, food processing, agriculture, enhanced oil recovery, and bioremediation.
A class of biosurfactants consisting of a sugar (glyco) moiety linked to a fatty acid (lipid) tail. The three main types are rhamnolipids (produced by Pseudomonas species), sophorolipids (produced by Starmerella bombicola), and mannosylerythritol lipids (MEL, produced by Pseudozyma/Moesziomyces species). Glycolipids are the most commercially advanced biosurfactants, with sophorolipids reaching industrial production and rhamnolipids available as commercial products (Evonik Zonix/Quix).
A glycolipid biosurfactant composed of one or two rhamnose sugars linked to β-hydroxy fatty acid chains. Traditionally produced by Pseudomonas aeruginosa, but non-pathogenic production in Pseudomonas putida (strain KT2440) is preferred for commercial applications. Rhamnolipids are excellent emulsifiers with applications in cleaning, agriculture, and bioremediation. Mono-rhamnolipids are particularly effective at disrupting biofilms, relevant to combating antimicrobial resistance (AMR). The only commercial rhamnolipid products are Evonik’s Zonix and Quix.
A glycolipid biosurfactant produced by the yeast Starmerella bombicola, consisting of a sophorose sugar linked to a hydroxyl fatty acid. Sophorolipids can exist in lactonic (ring) or acidic (open) form, with different surface-active properties. They are among the most commercially advanced biosurfactants due to high fermentation titers (>100 g/L achievable) and relatively simple downstream processing — in some processes, DSP accounts for as little as 1% of total production cost due to natural phase separation.
Augur implements every concept on this page — from PINNs to COGS modeling. Request access to see production-scale predictions from your own lab data.