Help of OSML

Figure 1. Flowchart of OSML: an on-site implemented predictor for protein-ligand binding sites prediction.

    Note that in OSML, the base dataset consisting of annotated ligand-binding protein sequences is hierarchically organized to facilitate the user to perform protein-ligand binding sites predictions on three different levels (refer to Fig. 1). More specifically, OSML provides three options, i.e., Level I (Regardless of Ligand Category and Type), Level II (Ligand Category), and Level III (Ligand Type), for user to designate on which level the query-driven dataset will be generated. For the convenience of the subsequent description, we call it a Level I Test if the query-driven dataset is generated on Level I of the base dataset. Similarly, Level II Test and Level III Test denote that the query-driven datasets are generated on Level II and III of the base dataset, respectively. More specifically, Level I Test, Level II Test, and Level III Test are defined as follows:
    Level III Test: If Level III option is chosen, the user should also designate the Ligand Type (e.g., ATP or Ca²⁺); then, OSML will generate query-driven dataset by performing PSI-BLAST with the subset, which consists of the corresponding annotated sequences that belongs to the designated Ligand Type, as base dataset.
    Level II Test: If Level II option is chosen, the user should also designate the Ligand Category (e.g., Nucleotide or Metal-Ion); then, OSML will generate query-driven dataset by performing PSI-BLAST with the combined set, which consists of all the annotated sequences that belong to the designated Ligand Category, as base dataset.
    Level I Test: If Level I option is chosen, OSML will generate query-driven dataset by performing PSI-BLAST with the combined set, which consists of all the annotated sequences (i.e., the entire base dataset) regardless of the Ligand Types and Ligand Categories they belong to, as base dataset.