Antibody Production Screening Example

DESCRIPTION:

This screening experiment was conducted to study how changing treatment conditions would affect the yield of monoclonal antibodies produced in mice innoculated with hybridoma cells. The response is proportional to the number of antibody molecules produced. The experimental design is a fractional factorial, ff0616, that has 6 experimental factors each at two levels. The design consists of 16 runs and is a resolution IV, 1/4 fraction of the full factorial.

See Chapter 2 of Haaland (1989) for a more complete description and for a follow-up response surface design. There is also a help file for the follow-up experiment; namely, abrsm.df.

ARGUMENTS:

RadDos
an experimental factor giving the radiation dose in rads.
Prime1
an experimental factor specifying the time in weeks between the initial injection of Pristane oil and the innoculation with antibody producing cells.
VolPrs
an experimental factor giving the volume of Pristane oil injected.
CelNum
an experimental factor showing the number of antibody producing cells used in the innoculation.
Growth
an experimental factor indicating the growth state of the antibody producing cells, either saturated or log stage.
Prime2
an experimental factor indicating whether or not a second priming with Pristane oil was used immediately prior to innoculation with the antibody producing cells
TtrVol
the response variable, antibody titer adjusted for volume. The measured value is proportional to the number of monoclonal antibody molecules produced. The response should be maximized.

REFERENCES:

Haaland, P. D. (1989). Experimental Design in Biotechnology. New York: Marcel Dekker, Chapter 2.

SOURCE:

Used by permission of Marcel Dekker, Inc.

SEE ALSO:

EXAMPLES:

# This design is already available in S-PLUS  
# under the name abscrn.df. These are the commands 
# that were used to create it: 
abscrn.fnames <- list(RadDos=c(250,500),Prime1=c(1,3), 
     VolPrs=c(0.1,0.5),CelNum=c('10e6','10e7'),Growth= 
     c('Log','Sat'),Prime2=c('No','Yes')) 
abscrn.design <- design.digest(rep(2,6),abscrn.fnames, 
     fraction=~RadDos:Prime1:VolPrs:Prime2 + 
     RadDos:CelNum:Growth:Prime2+ 
     Prime1:VolPrs:CelNum:Growth) 
abscrn.TtrVol <- c(70,150,34,32,138,56,123,225,50,2.7, 
                 1.2,12,90,2.1,4,15) 
abscrn.df <- cbind(abscrn.design,TtrVol=abscrn.TtrVol) 
# Sample analysis 
plot(abscrn.df) 
abscrn.fac <- fac.aov(abscrn.df) 
summary(abscrn.fac) 
pareto(abscrn.fac,sig=.10) 
abscrn2.df <- abscrn.df[1:8,-1] 
abscrn2.fac <- fac.aov(abscrn2.df) 
pareto(abscrn2.fac,sig=.1) 
qqnorm(abscrn2.fac,sig=.1)