Figure 4. Classification and clustering of the M. mairpaludis proteome. Each row represents an ORF from M. maripaludis. The annotation of each ORF is listed at the end of the row. The columns represent experimental conditions. NH4/H2, ammonia-limiting compared with H2-limiting growth conditions; P/H2, phosphate-limiting compared with H2-limiting growth conditions; P/NH4, phosphate-limiting compared with ammonia-limiting growth conditions. The abundance changes are shown in gray scale, with black showing no change in abundance. A color version of the figure can be found in the APPENDIX. (A) Flagella protein expression class. Proteins were classified as matching the abundance pattern of flagella proteins by means of the ‘K nearest neighbor’ algorithm in MEV (www.tm4.org/mev.html). A leave-one-out validation was conducted for the classification (Theilhaber et al., 2002). The procedure involves re-running the classification, leaving out one of the original training group proteins, and doing this for all of the members of the original training group. Proteins that were present in the class across the entire leave-one-out validation are indicated in bold. (B) A subsection of the agglomerative hierarchical clustering of the M. maripaludis proteome in MEV. The dendrogram shows the protein relationships in the clusters.