Semi-Supervised Threshold Queries on Pharmacogenomics Time Sequences

The analysis of time series data is of capital importance for pharmacogenomics since the experimental evaluations are usually based on observations of time dependent reactions or behaviors of organisms. Thus, data mining in time series databases is an important instrument towards understanding the effects of drugs on individuals. However, the complex nature of time series poses a big challenge for effective and efficient data mining. In this paper, we focus on the detection of temporal dependencies between different time series: we introduce the novel analysis concept of threshold queries and its semi-supervised extension which supports the parameter setting by applying training datasets. Basically, threshold queries report those time series exceeding an user-defined query threshold at certain time frames. For semi-supervised threshold queries the corresponding threshold is automatically adjusted to the characteristics of the data set, the training dataset, respectively. In order to support threshold queries efficiently, we present a new efficient access method which uses the fact that only partial information of the time series is required at query time. In an extensive experimental evaluation we demonstrate the performance of our solution and show that semi-supervised threshold queries applied to gene expression data are very worthwhile. Data mining in time series data is a key step within the study of drugs and their impact on living systems, including the discovery, design, usage, modes of action, and metabolism of chemically defined therapeutics and toxic agents. In particular, the analysis of time series data is of great practical importance for pharmacogenomics. Classical time series analysis is based on techniques for forecasting or for identifying patterns (e.g. trend analysis or seasonality). The similarity between time series, e.g. similar movements of time series, plays a key role for the analysis. In this paper, we introduce a novel but very important similarity query type which we call threshold query. Given a time series database DB, a query time series Q, and a query threshold τ , a threshold query TSQ DB(Q, τ ) returns those time series X ∈ DB having the most similar sequence of time intervals in which the time series values are above τ .I n other words, we assume that each time series X ∈ DB ∪{ Q} is transformed into a sequence of disjoint time intervals covering only those values of X that are (strictly) above the threshold τ . Then, a threshold query returns for a given query object Q that object X ∈ DB having the most similar sequence of time intervals. Let us note that the exact values of the time series are not considered, rather we are only interested in whether the time series is above or below a given threshold τ . In other words, the concept of threshold queries enables us to focus only on the duration of certain events indicated by increased time series amplitudes, while the degree of the corresponding amplitudes are ignored. This advantage is very beneficial, in particular, if we want to compare time