c

Table 5.3. Phase space dimension d, information coefficient i, and correlation coefficient p, corresponding to each data pair, and the result of the heuristic classification scheme. Classifications I, C and R stand for Independent, Coupled and Redundant, respectively.

Table 5.3. Phase space dimension d, information coefficient i, and correlation coefficient p, corresponding to each data pair, and the result of the heuristic classification scheme. Classifications I, C and R stand for Independent, Coupled and Redundant, respectively.

the coordinates of interest, we see that, 8 out of 15 pairs are classified as Coupled, and 7 of them are classified as Independent coordinates. The Coupled pairs, STZ, Sa, SV, XH, XS. Xj, 7a and fV expected to appear in the equations of motion of one-another. For instance, since S and 1Z are found to be Coupled, the dynamics of <S should be affected directly by the value of 71, or the dynamics of 7Z should be affected directly by the value of 5, or both. To be on the safe side, we assume the "both" case. Also, a coordinate in the phase space may appear in its own dynamics, so a generic form of equations of motion should also take this into account. On the other hand, the pairs TZa, TZj, 7ZV, Sj, Xa, XV and aP are classified to be Independent, and these pairs will not appear in the dynamics of each other.

Consequently, we propose that a suitable model for the fermentation behavior of recombinant Saccharomyces cerevisiae cells should be of the following form:

Writing down such a set of generic model equations for potential models reduces the computational effort of parameter estimation to about one-fourth, while increasing the reliability of the model constructed by increasing its degrees of freedom. In a modeling attempt with 100 data points, this corresponds to about a four-fold increase in reliability.

Was this article helpful?

## Post a comment