1) 若直接以20种氨基酸统计k_word: (以ZD98数据集为例) k Dimension 2 400 3 6490 4 22265 维数太大不适用构造特征向量 考虑氨基酸约化后特征提取 约化方案: Classification Abbreviation Abbreviation Strongly hydrophilic or polar L R, D, E, N, Q, K, H Strongly hydrophobic B L, I, V, A, M, F Weakly hydrophi