Meeting Abstract

90.6  Monday, Jan. 6 14:45  Computational methods to determine the instantaneous respiratory patterns of animals from respirometry data PENDAR, H*; SOCHA, JJ; Virginia Tech; Virginia Tech hpendar@vt.edu

Flow-through respirometry is a commonly used technique to record gas exchange patterns of CO2 and O2. This powerful technique has yielded great insight into animal physiology, but the true gas exchange that occurs by the animal is not recorded due to experimental limitations. This well-known problem results from both geometric and scale effects of experimental system and the gas flow rate. In order to draw time-accurate interpretations from the data or to sync it with other signals, gas exchange data must be transformed to instantaneous data. Mathematically, the recorded pattern of gas exchange is a convolution of the actual gas exchange of the animal and a transfer function. Depending on the transfer function, convolution could strongly change the actual pattern by smoothing the high frequency components, stretching the individual pulses, and combining the pulses together. These problems are more pronounced in small animals, which exhibit faster respiratory dynamics. This study presents two new methods to deconvolve respirometry data to reveal the instantaneous gas exchange patterns of animals. In the first, we use a stochastic approximation method, which is more accurate than previous methods and robust against the level of noise. In addition, it is not required to have prior knowledge about the noise. In the second method, we provide a novel but simple algorithm to find the transient CO2/O2 concentration point by point throughout the dataset. In contrast to other methods, this algorithm places no restriction on the number of data points. These new methods, which are validated with experimental data from insects, are broadly applicable for the quantification of physiological processes such as hormone secretion, insulin dynamics, and hepatic glucose production. Supported by NSF 0938047.