The problem of finding interesting sequential pattens from spatial or/and temporal sequence data has been intensively studied in data mining community. When we use sequence mining for scientific studies in medical or biological fields, it is desirable to be able to quantify the statistical significance of the discovered patterns in the form of p-values. Unfortunately, however, since pattern mining algorithms are designed to select the most interesting patterns from huge number of candidate patterns, we need to correct the selection bias (cf. multiple testing bias) when we evaluate the statistical significances of the discovered patterns. In this talk, we introduce two approaches for finding sequential patterns along with their statistical significances by borrowing idea from LAMP (Terada et al., PNAS 2015) and Selective Inference (Suzumura et al., ICML2017), and demonstrate their usefulness by applying them to animal trajectory data analysis.