Question

In a drug study, the company X desires to measure protein levels in samples derived from treated animals. There are 5 animals in each branch of the study. The sample processing is performed as follows: in each branch, the 5 animals are sacrificed, the tissue samples are derived from each animal and processed, then the processed preps are pooled together, and then 3 separate complete runs of mass-spectrometry measurements are performed on each pool, in the hope of reducing the noise/variation in the measured protein levels. At the end, the dataset consists of "samples" of those 3 independent protein level measurements for large number of proteins.

Think about this experimental design and explain if it is adequate or if it is missing something. If you find problems with this design, be specific: explain what is wrong and why, and what problems can this cause with downstream data processing and interpretation.

Answer 1

Although all proper care has been taken to avoid any experimental error, it seems that this experiment will be more inclined towards giving accurate results, not precise. This is because all the samples being investigated in the mass spectrophotometry procedure are already pooled. Thus, inter-animal differences would not be taken into account. In order to obtain both precise and accurate results, it would be wise to not pool all the samples and run spectrum analysis thrice but to run spectrum analysis thrice in three random animals out of all and then match the results. The average of all results can be then taken as final experimental value.

This change would bring both, precision and accuracy to the experiment.