Abstract:To better understand the dynamics of the underlying processes in cells, it is necessary to take measurements over a time course. Modern high-throughput technologies are often used for this purpose to measure the behavior of cell products like metabolites, peptides, proteins, [Formula: see text]RNA or mRNA at different points in time. Compared to classical time series, the number of time points is usually very limited and the measurements are taken at irregular time intervals. The main reasons for this are the costs of the experiments and the fact that the dynamic behavior usually shows a strong reaction and fast changes shortly after a stimulus and then slowly converges to a certain stable state. Another reason might simply be missing values. It is common to repeat the experiments and to have replicates in order to carry out a more reliable analysis. The ideal assumptions that the initial stimulus really started exactly at the same time for all replicates and that the replicates are perfectly synchronized are seldom satisfied. Therefore, there is a need to first adjust or align the time-resolved data before further analysis is carried out. Dynamic time warping (DTW) is considered as one of the common alignment techniques for time series data with equidistant time points. In this paper, we modified the DTW algorithm so that it can align sequences with measurements at different, non-equidistant time points with large gaps in between. This type of data is usually known as time-resolved data characterized by irregular time intervals between measurements as well as non-identical time points for different replicates. This new algorithm can be easily used to align time-resolved data from high-throughput experiments and to come across existing problems such as time scarcity and existing noise in the measurements. We propose a modified method of DTW to adapt requirements imposed by time-resolved data by use of monotone cubic interpolation splines. Our presented approach provides a nonlinear alignment of two sequences that neither need to have equi-distant time points nor measurements at identical time points. The proposed method is evaluated with artificial as well as real data. The software is available as an R package tra (Time-Resolved data Alignment) which is freely available at: http://public.ostfalia.de/klawonn/tra.zip .

[Acute myocardial infarction in a 13-year-old boy with Duchenne's disease].

OAW-GAN: Occlusion-Aware Warping GAN for Unified Human Video Synthesis

Multilevel Dynamic Time Warping: A Parameter-Light Method for Fast Time Series Classification

Multiscale Manifold Warping

A Global Approach for Video Matching

Analysis of temporal alignment for Video Classification

A General Optimization Framework for Dynamic Time Warping

Temporal Alignment of Human Motion Data: A Geometric Point of View

Alignment of time-resolved data from high throughput experiments

shapeDTW: shape Dynamic Time Warping

Affine and Regional Dynamic Time Warpng

Dynamic State Warping

Sparsification of the Alignment Path Search Space in Dynamic Time Warping

Spatio-Temporal Alignments: Optimal transport through space and time

Diffeomorphic Transformations for Time Series Analysis: An Efficient Approach to Nonlinear Warping

TimewarpVAE: Simultaneous Time-Warping and Representation Learning of Trajectories

Efficient Processing of Warping Time Series Join of Motion Capture Data

Video Face Editing Using Temporal-Spatial-Smooth Warping

ESDTW: Extrema-based shape dynamic time warping

mnmDTW: An extension to Dynamic Time Warping for Camera-based Movement Error Localization

Dynamic Time Warping: Itakura vs Sakoe-Chiba