Ordered Bijective Interpolated Warping (OBI-Warp) aligns matrices along a single axis using Dynamic Time Warping (DTW) and a one-to-one (bijective) interpolated warp function. OBI-Warp harnesses the non-linear, comprehensive alignment power of DTW and builds on the discrete, non-bijective output of DTW to give natural interpolants that can be used across multiple datasets.
OBI-Warp was developed specifically for the chromatographic alignment of complex mass spectrometry (MS) proteomics data. Using high confidence MS/MS identifications as time standards, OBI-Warp default parameters have been optimized to give accurate alignments under a variety of real-world conditions including datasets with little overlapping signal. Command-line options to override defaults are available (e.g., gap penalty, local weights and percent of bijective anchors). Though developed for MS proteomics data, OBI-Warp is suited to a wide variety of alignment problems.
Pearson's correlation coefficient, covariance, dot product, and Euclidean distance have been implemented as the available vector similarity functions. Redundant calculations for correlation coefficient and covariance are cached in the n x m comparisons to give the algorithmic equivalent of calculating the dot product.