Glioma is a complex malignant neoplasm, treating each case as a dark box obviously could not acquire expected overall prognosis, because subpopulation inside tumor could be core factor which directly influence tumor progression. So, clonal sequencing or single-cell sequencing could be best solution to this problem. Here we combined these two strategies and analyzed more than 6000 single-cell transcriptome of 73 surgery points from 13 glioma and 1 brain metastasis patient. After these cells assigned into normal, immune, tumor and several other groups based on differentially expressed genes and specified TCGA 4 classification gene sets, spacial information of sampling points helped us built up cell type constituent of different part inside tumor, and illustrated invasion patterns during tumor growth.