Meeting Abstract
24.6 Friday, Jan. 4 Problem and solution: Multiplexing distorts metabolic data LIGHTON, JRB*; FOERSTER, TD; KAIYALA, KJ; WISSE, B; Sable Systems International; Sable Systems International; University of Washington; University of Washington lighton@sablesys.com
When measuring the metabolic rates of multiple animals, it is common practice to sample excurrent air from each cage or chamber, and direct these air samples to a single gas analyzer chain. These samples are analyzed in succession (or “multiplexed”), interleaved with periodic analysis of incurrent air composition in order to compensate for analyzer drift and fluctuations in incurrent gas concentrations. Each such analysis takes a finite time, and must be completed before the next sample is analyzed. Thus an appreciable time – the “cycle time” of the system – will elapse between successive measurements of a given animal. The actual metabolic signal from each animal is therefore composed of a series “metabolic snapshots” which are separated by the cycle time of the system. This approach has the advantage of requiring only a single gas analyzer chain, and thus lowering costs. However, it suffers from two major disadvantages. First, rapidly changing metabolic signals may be missed, or, even worse, distorted by aliasing effects. Second, the nature of the sampled data depends critically on the moment at which the sampling cycle is initiated. Because the effect of starting time cannot be predicted because its effects lie along the future path of time’s arrow, the results of any multiplexing system include a strong stochastic component, especially where the metabolic data are variable. As a result, resting energy expenditure (REE) is generally overestimated and activity EE (AEE) is underestimated. Using a Promethion multiple-animal, continuous (non-multiplexed) metabolic phenotyping system, we model a variety of multiplexed systems using continuous data from 16 mice, demonstrating and quantifying the serious errors that result from multiplexing.
