Improvement for Missing Value Imputation in Microarray Data
C.Yang Andy · 2009
Prediction of regulatory genes via amouns of microarray time-series data is an important issue for drug development and disease research. Biologists can retrieve lots of information about genes according to reactions to specific probes on each microarray chip. Several statistical approaches are commonly used to predict gene regulations such as clustering and similarity analysis. However, even genes with known regulations may have reaction delay or offsets among time axis in microarray experiment results. Moreover , certain microarray experiment results show that outliers exist at specific time slot point. Therefore, it is not suitable to calculate gene similarity with whole microarray time-series data. In this paper we propose a novel approach combining K-Nearest Neighbor(KNN) and Dynamic Time Warping(DTW), along with several techniques for accuracy improvement for DTW algorithm. With the modifications for improvement of DTW, missing value imputation can hence be more accurate.