engleski

Normalization and batch correction methods for high-throughput glycomics

Glycomics is rapidly emerging field in high-throughput biology that aims to systematically study glycan structures of a given protein, cell type or organic system. As it is characteristic for any other high-throughput method in biology (microarrays, next-generation sequencing, mass spectrometry), accuracy of high-throughput glycomics methods is highly affected by complicated experimental procedure. Standard experiment requires highly trained personnel, complicated sample collection and preparation procedure, large set of chemicals and calibrated machines. Standard study includes 1000 to 2000 samples, experiment can take several months and during that time many experimental conditions can vary. As a consequence, differences in experimental procedure represent huge source of variation and need for normalization and batch correction arises naturally. We compared most popular normalization and batch correction methods, from microarray and metabolomics field, on several glycomics datasets. We evaluated them based on variation of standards and correlation of replicates. According to standard variation and replicate correlation measures, every normalization and batch correction method performs relatively well, showing that use of any preprocessing method decreases experimental variation and increases the statistical power of the analysis.

batch effects; normalization; noise; glycomics

nije evidentirano