To analyze large data sets for `human-size' information, we need data`distillation' to reduce the dimensionality. We devised a non-metric MDS (nMDS) solely based on rank orders, and have been applying it and related methods to bioinformatics-related data. The purpose of this informal talk is two-fold: 1) to advertise our nMDS and 2) to ask for specialists' advices and (hopefully) help. After outlining our nMDS and illustrations, I will exhibit biological examples analyzed by nMDS. Then, I advocate a general method (called data honing) to remove `outlying points' in pattern extraction / dimension-reduction methods with some illustrations.