ANÁLISIS DE LA RELACIÓN DE LAS VARIABLES GÉNERO, EDAD, NIVEL DE

In recent years there has been increasing interest in the application o f simulation, and computer techniques in general, to improve the design and operation o f protein downstream processes. This has come about because the downstream process often forms the major cost in the production o f a protein and its optimisation therefore represents a way for biotechnology companies to maintain a competitive advantage (Wheelwright,

1986). However, the bioprocess industry has traditionally used empirically based methods for the design o f downstream processes, such as the scale-up o f laboratory processes via a pilot-plant, and relied on labour-intensive techniques during protein manufacture which are often time consuming and/or expensive. The large number o f alternatives currently available for each stage o f a downstream process (Section 1.2), the increasing use o f recombinant systems (Section 1.3) and commercial pressures which call for a bioprocess to be developed in a shorter time and at lower cost have also added to the difficulty o f producing an optimal downstream process.

Attempts to rationalise the design o f protein downstream processes have been made by Bonneijea et al. (1986), Wheelwright (1986), Titchener-Hooker et al. (1991) and Leser and Asenjo (1992). This has led to a consolidation o f the knowledge required for design and opened the way for an increasing use o f computer tools in this area. Leser and Asenjo (1992) concluded that expert systems could be very useful for the synthesis o f protein downstream processes and their application has been examined by Stephanopoulos and Stephanopoulos (1986), Siletti (1988), Asenjo et al. (1989) and Turner et al. (1994). At the synthesis stage o f design, simulation can also be used to choose between alternative operations (Leser and Asenjo, 1992) and to examine the effect o f changes in the sequence o f operations using case studies.

Once synthesis is completed and a downstream processing sequence has been proposed, simulations can be used to assess the feasibility o f the process by predicting product purity and yield, calculating required equipment sizes and estimating capital and operating costs. Optimisation o f the process is also possible through the use o f case studies or numeric techniques such as successive quadratic programming (SQP) and the effect o f interactions within the process can be examined by simulating multiple operations.

Published work on the application o f simulation to the design o f protein downstream processes is reviewed below. The general purpose dynamic simulator SPEEDUP has been used to carry out simulations o f downstream processes by:

• Gritsis and Titchener-Hooker (1989), who optimised a protein processing sequence consisting o f ultrafiltration, precipitation and centrifugation using SQP to give the minimum processing time.

• Middelberg et al. (1989), who simulated the centrifugal recovery o f porcine growth hormone inclusion bodies to find the best operating conditions for inclusion body separation. Middelberg et a l (1992a) also simulated interactions between the fermentation stage and subsequent homogenisation and debris removal stages used for inclusion body recovery. While Middelberg (1995b) used simulation to optimise protein refolding.

• Siddiqi et a l (1991), who simulated interactions between homogenisation and centrifugation and used a case study to demonstrate how conditions for efficient protein release and recovery can be optimised.

• Clarkson et a l (1992, 1996a, 1996b), who simulated fractional precipitation and centrifugation operations in the recovery o f the intracellular enzyme alcohol dehydrogenase from bakers’ yeast and carried out a case study to demonstrate how the recovery o f precipitate particles could be improved. The importance o f pilot-scale model verification was highlighted.

• Bogle et a l (1991, 1993), who simulated the production and recovery o f porcine somatotropin inclusion bodies from Escherichia coli to optimise separation efficiency. Lu et a l (1994) have also used the general purpose dynamic simulator gPROMS to simulate a typical downstream processing sequence for the production o f intracellular enzymes. Wai et a l (1996) have used the dynamic simulator DYNSIM as part o f an integrated approach to bioprocess design, with flowsheet simulations being used to test process feasibility and perform overall process optimisation. The model system under consideration was the recovery o f bovine somatotropin, produced as an inclusion body by a recombinant E. coli strain. Samsatli and Shah (1996) have used a two stage approach to dynamically optimise the operation and then the scheduling o f a typical intracellular enzyme recovery process. While Zhou et a l (1997) simulated a process consisting o f a bakers' yeast fed-batch fermentation followed by cell harvesting, high pressure homogenisation and debris removal stages in order to examine process interactions using 'windows o f operation', a graphical representation o f the operational space determined by the system and engineering constraints (Woodley and Titchener-Hooker, 1996).

The chemical process simulator ASPEN-Plus has been used by Cooney et al. (1988) to simulate a downstream processing sequence consisting o f vacuum filtration, ultrafiltration and spray drying for the recovery o f a proteolytic enzyme. The effect o f various operating conditions on the process economics was then examined, although no verification o f user defined models was completed (Chapter 2.0).

While the application o f Aspen Tech's BioProcess Simulator (BPS) to the design o f downstream processes has been demonstrated by:

• Petrides et a l (1989), who simulated the production o f porcine growth hormone (with the help o f performance characteristics derived from laboratory and pilot-plant data) to perform simplified material balances and examine the economic feasibility o f the process

• Healy et al. (1990), who used simulation as part o f the development and design o f a multi-product enzyme production plant

• Bhattacharya (1993), who simulated gel permeation chromatography for insulin separation and used case studies to examine the sensitivity o f the process to elution flowrate.

BioProDesigner has been used by Petrides et al. (1995), who simulated a full production process for human insulin in order to enable equipment specification, calculate material balances, examine process scheduling and complete an economic and sensitivity analysis o f the process. Most of the above work which has used the general purpose simulation packages SPEEDUP and gPROMS has focused on unit integration and/or optimisation and only part of the full downstream process has been simulated. On the other hand, the work with BPS and BioProDesigner has concentrated on analysing the economics and/or sensitivity o f the full downstream process after steady-state material balances have been calculated.

The introduction o f automation to protein manufacture, often incorporating modelling and simulation, is another way o f producing a more optimal downstream process (Kossik and Miller, 1993). This involves utilising computer technology to add continuous process

monitoring, on-line measurement, automatic control and electronic batch scheduling to downstream processes, thereby making them semi-continuous (Ransohoff et a l, 1990). The benefits include increased productivity, lower operating costs, easier validation, better process control and improved safety. As in the chemical industry, the ability to perform real time dynamic simulation should also enable simulations to be used in the design and testing o f control schemes, the development o f operating procedures, operator training, trouble shooting during production and on-line optimisation (refer to Section 1.4.2). However, one obstacle which may need to be overcome before these applications are realised for protein downstream processes is the current lack o f rigorous, predictive models (Chapter 2.0).