Figure 1.
Schematic illustration of the MinPath method.
Assume 6 families (or orthologous groups, f1, …, f6) are identified from a given sample of genes (e.g., the genes could be from a genome, or sampled from a metagenome). The naïve mapping approach (shown on the left) will lead to a reconstruction with 4 pathways annotated (p1, p2, p3, and p4). Due to the overlapping nature of the biological pathways (see text for more details), pathway p3 shares function f3 with pathway p2. We claim that only three pathways, p1, p2, and p3 are sufficient to explain the existence of the 6 families annotated in the dataset, and a conservative reconstruction of pathways should have only 3 pathways (shown on the right). As we show in the paper, such a conservative estimation of pathways provides a more reliable estimation of the functional diversity of a sample.
Figure 2.
Comparison of the number of pathways reconstructed for various genomes by different methods.
The coloring schema is as following: MinPath (red triangles), naïve mapping approach (green), and the pathway annotation maintained in KEGG database after human evaluation (blue).
Figure 3.
The ascorbate and aldarate metabolism pathway, eliminated by MinPath.
The diagram was prepared based on the corresponding KEGG pathway (ID = 00053), and only part of the pathway is shown for clarity. The three enzymes that are annotated in the human genome are highlighted in green, even though none of these enzymes are unique to this pathway.
Table 1.
Selected spurious pathways of the human genome that are incorrectly identified by the naïve mapping approach.
Table 2.
Selected spurious pathways of the E. coli genome (collected in KEGG) eliminated by MinPath.
Table 3.
Comparison of biological pathway reconstruction based on MinPath and the naïve mapping approach for selected metagenomesa.