Definition of Representative Set

[NatureGeorge]

# pip install pdb-profiling==0.3.4a6 # NOTE: (current version)
from pdb_profiling import default_config
from pdb_profiling.processors import SIFTS

default_config(your_workdir)

to_display_cols = ['UniProt', 'new_identity', 'pdb_id', 'chain_id', 'new_unp_range',
                   'select_rank', 'oc_cluster', 'rmsd_cluster', 'oc_select_tag', 'oc_rmsd_select_tag']
demo_for_diff_coverage = SIFTS('P43220').pipe_cluster3d_mo().result()
demo_for_diff_coverage_conformers = SIFTS('P21359-2').pipe_cluster3d_mo().result()

• new_unp_range: the pdb chain's coverage region of the target UniProt isoform
• select_rank: rank derived from sorting criteria (default: ['bs_score', '1/resolution', 'revision_date', 'id_score'])
• oc_cluster: overlap coefficient based coverage (i.e. segment) cluster, each number represent a cluster
  • greedy algorithm: follow the select_rank, iteratively compare the candidate chain with those chains in an existing oc_cluster, if not meet the cutoff with any existing oc_clusters, define a new oc_cluster
  • default cutoff for overlap coefficient: 0.8, i.e. include the chain into the cluster if its OC is great than or equal to 0.8 when compared with any element in that cluster
• rmsd_cluster: R.M.S.D based conformer cluster within a coverage cluster
  • greedy algorithm: within each oc_cluster, follow the select_rank, iteratively compare the candidate chain with those chains in an existing rmsd_cluster, if not meet the cutoff with any existing rmsd_clusters, define a new rmsd_cluster
  • default cutoff for R.M.S.D: 2.5, i.e. include the chain into the cluster if its R.M.S.D is less than or equal to 2.5 when compared with any element in that cluster
• oc_select_tag: whether to include the chain into representative set that consider coverage only
  • select the chain with the best rank of each oc_cluster
• oc_rmsd_select_tag: whether to include the chain into the representative set that considers both coverage and conformers
  • select the chain with the best rank of each rmsd_cluster in the context of corresponding oc_cluster

demo_for_diff_coverage[to_display_cols]

demo_for_diff_coverage_conformers[to_display_cols]

---

Switch to SIFTS('P43220').pipe_select_mo().result() when you only need to consider coverage's overlap coefficient and do not care about R.M.S.D.