Data volumes collected in many scientific fields have long exceeded the capacity of human comprehension. This is especially true in biomedical research where multiple replicates and techniques are required to conduct reliable studies. Ever-increasing data rates from new instruments compound our dependence on statistics to make sense of the numbers. The currently available data analysis tools lack user-friendliness, various capabilities or ease of access. Problem-specific software or scripts freely available in supplementary materials or research lab websites are often highly specialized, no longer functional, or simply too hard to use. Commercial software limits access and reproducibility, and is often unable to follow quickly changing, cutting-edge research demands. Finally, as machine learning techniques penetrate data analysis pipelines of the natural sciences, we see the growing demand for user-friendly and flexible tools to fuse machine learning with spectroscopy datasets. In our opinion, open-source software with strong community engagement is the way forward. To counter these problems, we develop Quasar, an open-source and user-friendly software, as a solution to these challenges. Here, we present case studies to highlight some Quasar features analyzing infrared spectroscopy data using various machine learning techniques.

We compared hyperspectral infrared raster maps and images for contrast, definition and resolution of the same samples recorded with a confocal microscope coupled with a synchrotron radiation source vs a Focal Plane Array (FPA) detector equipped microscope. Biological samples (hair and skin sections) and astrophysics samples (meteoritic grains) were used. The samples presented are a few microns in size, such as embedded particles, a single unique cell or thin layer. Our results show that the actual spatial resolution and contrast of FPA images were lower than spectral maps from the confocal microscope. The FPA microscope also produced measurements that lacked accuracy: size of sample features and peak intensity were inaccurately estimated. More surprisingly, the intensity of absorption peaks in the FPA images was lower than the intensity measured from the same sample with a confocal microscope. Our measurements underlined the complementarity of FPA and confocal microscopes. FPA can be used to quickly measure the overall composition of a sample and detect the distribution of its components, but may fail measuring the exact chemical composition of the small features and may not detect weak spectral differences between adjacent positions. The averaging effect of aperture-less systems not only affects image resolution but also lowers their spectral accuracy. Confocal microscopes are inherently slower but give a more accurate measurement of the local composition at the diffraction limit.

Background: Computational methods that infer single nucleotide polymorphism (SNP) interactions from phenotype data may uncover new biological mechanisms in non-Mendelian diseases. However, practical aspects of such analysis face many problems. Present experimental studies typically use SNP arrays with hundreds of thousands of SNPs but record only hundreds of samples. Candidate SNP pairs inferred by interaction analysis may include a high proportion of false positives. Recently, Gayan et al. (2008) proposed to reduce the number of false positives by combining results of interaction analysis performed on subsets of data (replication groups), rather than analyzing the entire data set directly. If performing as hypothesized, replication groups scoring could improve interaction analysis and also any type of feature ranking and selection procedure in systems biology. Because Gayan et al. do not compare their approach to the standard interaction analysis techniques, we here investigate if replication groups indeed reduce the number of reported false positive interactions. Results: A set of simulated and false interaction-imputed experimental SNP data sets were used to compare the inference of SNP-SNP interactions by means of replication groups to the standard approach where the entire data set was directly used to score all candidate SNP pairs. In all our experiments, the inference of interactions from the entire data set (e.g. without using the replication groups) reported fewer false positives. Conclusions: With respect to the direct scoring approach the utility of replication groups does not reduce false positive rates, and may, depending on the data set, often perform worse.