GCalignR: An R package for aligning gas-chromatography data for ecological and evolutionary studies

Abstract

Chemical cues are arguably the most fundamental means of animal communication and play an important role in mate choice and kin recognition. Consequently, there is growing interest in the use of gas chromatography (GC) to investigate the chemical basis of eco-evolutionary interactions. Both GC-MS (mass spectrometry) and FID (flame ionization detection) are commonly used to characterise the chemical composition of biological samples such as skin swabs. The resulting chromatograms comprise peaks that are separated according to their retention times and which represent different substances. Across chromatograms of different samples, homologous substances are expected to elute at similar retention times. However, random and often unavoidable experimental variation introduces noise, making the alignment of homologous peaks challenging, particularly with GC-FID data where mass spectral data are lacking. Here we present , a user-friendly R package for aligning GC-FID data based on retention times. The package was developed specifically for ecological and evolutionary studies that seek to investigate similarity patterns across multiple and often highly variable biological samples, for example representing different sexes, age classes or reproductive stages. The package also implements dynamic visualisations to facilitate inspection and fine-tuning of the resulting alignments and can be integrated within a broader workflow in R to facilitate downstream multivariate analyses. We demonstrate an example workflow using empirical data from Antarctic fur seals and explore the impact of user-defined parameter values by calculating alignment error rates for multiple datasets. The resulting alignments had low error rates for most of the explored parameter space and we could also show that performed equally well or better than other available software. We hope that will help to simplify the processing of chemical datasets and improve the standardization and reproducibility of chemical analyses in studies of animal chemical communication and related fields.

Chemical patterns of colony membership and mother-offspring similarity in Antarctic fur seals are reproducible over time

“…For example, because replication studies often produce weaker effect sizes than original studies (Simonsohn, 2015; Open science collaboration, 2015), we attempted to enlarge our sample size of mother-offspring pairs as far as was practicable. We also improved the standardization and reproducibility of our chemical analysis pipeline by performing peak detection with open source software and by integrating the alignment algorithm of Stoffel et al (2015) into an R package (Ottensmann et al, 2018). However, these minor modifications appear to have been of little consequence as the effect sizes of colony membership and mother-offspring similarity did not differ systematically between the two studies.…”

“…The resulting GC-MS data were then processed using OpenChrom (Wenig & Odermatt, 2010) for detection and correction of split peaks. Afterwards, we used GCalignR in R (Ottensmann et al, 2018; R Core Team, 2019) to align the resulting chromatograms by correcting minor shifts in retention times among samples and maximizing the number of shared components.…”

Peak alignment of gas chromatography–mass spectrometry data with deep learning

“…We compare our alignment results with two existing algorithms: the correlation optimized warping (COW) algorithm [27,28,29] 5 , and the GCalignR algorithm [15]. We investigated many other existing algorithms, however, most are unsuitable to compare with our algorithm -in some of the exiting algorithms the peak alignment module was not designed to be used on its own [30,11]; some algorithms do not have source code available [12]; some requires a specific MATLAB toolbox that we do not own [14,10]; and another built on proprietary software [13].…”

“…These authors concluded that due to the complexity of the metabolome, all existing software will require further improvement and researchers are recommended to perform manual checks on the alignment of important biomarkers. More alignment algorithms have been developed since the review papers above [10,11,12,13,14,15], many of these utilising the mass spectral information for alignment. Current alignment algorithms are all based on traditional, symbolic artificial intelligence (AI) techniques, that is, the use of a set of formal, mathematical rules.…”

Genetic divergence and phenotypic plasticity contribute to variation in cuticular hydrocarbons in the seaweed flyCoelopa frigida

“…The total ion chromatograms were quality checked, de-noised and Savitzky-Golay 146 filtered before peak detection and peak area integration in OpenChrom®. Peaks were 147 then aligned using the R-package 'GCalignR' (Ottensmann et al 2018) prior to statistical 148 analysis. Peaks were tentatively identified by their mass spectra and retention time index 149 based on a C21-C40 n-alkane standard solution (Sigma-Aldrich).…”