Back to TaskOperations page.

AddCDRProfileSetsOperation

Author

Jared Adolf-Bryfogle; jadolfbr@gmail.com; PI: Roland Dunbrack

Part of the RosettaAntibody and RosettaAntibodyDesign (RAbD) Framework

Brief

Add Cluster-based sets of mutations as a TaskOperation. Essentially samples full sequences of CDRs within a particular CDR cluster randomly each time the packer is called. Does this for each CDR. Uses the MutationSetDesignOperation for the heavy lifting (not currently RS compatible) If a CDR has an unknown cluster or there are no data for that particular CDR, will skip that CDR. CDR definitions used are North/Dunbrack as the clusters are defined using it.

Details

Note that by default, a data cutoff of 10 is set. If the cluster has less than 10 sequences it will be skipped. Use the set_cutoff function to change this.

This TaskOperation is not currently recommended for H3 as it does not cluster well

<AddCDRProfileSetsOperation cdrs="(&string,&string)" numbering_scheme="(&string)" include_native_restype="(&bool, true)" picking_rounds="(&size, 1)"/>

Common Options

  • cdrs (&string,&string) (default=all cdrs): Select the set of CDRs you wish to restrict to (ex: H1 or h1)
  • numbering_scheme (&string): Set the antibody numbering scheme. This option can also be set through the command line. See General Antibody Tips for more info.
  • include_native_restype (&bool) (default=true): Include the native residue type when sampling?
  • picking_rounds (&size) (default=1): Set the number of times a residutype is chosen from the distribution each time the packer is generated. Increase this number to increase variability of design.

Length-based Sampling Options

  • limit_only_to_length (&bool) (default=false): Set the class to sample all CDR sequences within a particular length, instead of by cluster, which is the default. If the length is small and the clusters are close together, this would work OK. Generally not recommended though.
  • use_light_chain_type (&bool) (default=true): Use the light chain type if passed using command line options (-light_chain [lambda/kappa]). Recommended when sampling only sequences for particular lengths.

Uncommon Options

  • use_outliers (&bool) (default=false): Use cluster outliers as defined using DihedralDistance and RMSD.
  • cutoff (&size) (default=10): Will use the fallback strategy for this CDR if the total is less than or equal to this number.

Benchmarking Options

  • force_north_paper_db (&bool) (default=false): Force the use of the original 2011 North/Dunbrack clustering paper data as the database instead of any up-to-date versions downloaded from PyIgClassify.

See Also