Thumbnails:
List:
Year:
Category:
Session:
Poster:
Getting poster data...
Jelena Čuklina, Chloe Lee, Evan G. Williams, Tatjana Sajic, Ben Collins, María Rodríguez Martínez, Ruedi Aebersold (Otto-Stern-Weg 3, Zürich, Switzerland; Ph.D. Program in Systems Biology, Zurich, Switzerland; IBM Research, Rüschlikon, Switzerland; University of Zurich, Switzerland)
Technical advances in high-throughput technologies such as mass spectrometry and NGS have increased sample throughput to a degree that large-scale studies consisting of hundreds of samples are becoming routine. However, large scale datasets suffer from bias, known as “batch effects”, that distort signal in the data and decrease the power to identify biological variance. While batch effects are common to all “omics”, each method has specific issues. We present proBatch - a novel R package that facilitates visualization and correction of batch effects. The package unifies the interface to commonly used visualization techniques such as boxplots, hierarchical clustering and Principal Variance Component Analysis. To correct the bias in the data, we implement several normalization methods. Additionally, we introduce a new technique for correction of intensity drifts, specific for MS data. This technique fits a LOESS curve to adjust for continuous drift. Finally, we provide functions for quality control and efficacy of the applied correction, for example, replicate correlation comparison. The package is available for installation: https://github.com/symbioticMe/proBatch.