Accessing many fundamental questions in biology begins with empirical estimation of simple monotonic rates of underlying biological processes. Across a variety of disciplines, ranging from physiology to biogeochemistry, these rates are routinely estimated from non-linear and noisy time series data using linear regression and ad hoc manual truncation of non-linearities. Here, we introduce the R package LoLinR, a flexible toolkit to implement local linear regression techniques to objectively and reproducibly estimate monotonic biological rates from non-linear time series data, and demonstrate possible applications using metabolic rate data. LoLinR provides methods to easily and reliably estimate monotonic rates from time series data in a way that is statistically robust, facilitates reproducible research and is applicable to a wide variety of research disciplines in the biological sciences.
The authors declare no competing or financial interests.
Conceptualization: C.O. and D.R.B.; Methodology: C.O. and D.R.B.; Software: C.O. and D.R.B.; Writing – original draft preparation: C.O.; Writing – review and editing: C.R.W., D.J.M. and D.R.B.; Funding acquisition: D.J.M. and C.R.W.
This work was funded by a Monash University Dean's International Postgraduate Student Scholarship to C.O., Australian Research Council grants to D.J.M. and C.R.W., and a Monash University Centre for Geometric Biology Post-Doctoral Fellowship to D.R.B.
All data used in this study are included in the LoLinR package and available at https://github.com/colin-olito/LoLinR.
Supplementary information available online at http://jeb.biologists.org/lookup/doi/10.1242/jeb.148775.supplemental
- Received August 26, 2016.
- Accepted December 14, 2016.
- © 2017. Published by The Company of Biologists Ltd