Last Doc Update: 6/25/2020


MetaData

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

Rosetta Antibody Design (RAbD): A General Framework for Computational Antibody Design, PLOS Computational Biology, 4/27/2018

Jared Adolf-Bryfogle, Oleks Kalyuzhniy, Michael Kubitz, Brian D. Weitzner, Xiaozhen Hu, Yumiko Adachi, William R. Schief, Roland L. Dunbrack Jr.


Overview

RosettaAntibodyDesign (RAbD) is a generalized framework for the design of antibodies, in which a user can easily tailor the run to their project needs. The algorithm is meant to sample the diverse sequence, structure, and binding space of an antibody-antigen complex. It can be used for a multitude of project types, from denovo design to redesigns that improve binding affinity, optimize stability, or manipulate function.

The framework is based on rigorous bioinformatic analysis and rooted very much on our recent clustering of antibody CDR regions. It uses the North/Dunbrack CDR definition as outlined in the North/Dunbrack clustering paper.

The supplemental methods section of the published paper has all details of the RosettaAntibodyDesign method. This manual serves to get you started running RAbD in typical use fashions.

A PyRosetta Notebook for RAbD and framework components can be found here

The preprint of the PyRosetta Notebook Workshop is available here

The beginning of this manual reads like a tutorial, with more advanced concepts to RAbD being introduced in each example.

Algorithm

Broadly, the RAbD protocol consists of alternating outer and inner Monte Carlo cycles. Each outer cycle consists of randomly choosing a CDR (L1, L2, etc…) from those CDRs set to design, randomly choosing a cluster and then a structure from that cluster from the database according to the input instructions, and grafting that CDR’s structure, onto the antibody framework in place of the existing CDR (GraftDesign). The program then performs N rounds of the inner cycle, consisting of sequence design (SeqDesign), energy minimization, and optional docking. Each inner cycle structurally optimizes the backbone and repacks side chains of the CDR chosen in the outer cycle as well as optional neighbors in order to optimize interactions of the CDR with the antigen and other CDRs.

Backbone dihedral angle (CircularHarmonic) constraints derived from the cluster data are applied to each CDR to limit deleterious structural perturbations. Amino acid changes are typically sampled from profiles derived for each CDR cluster in PyIgClassify. Conservative amino acid substitutions (according to the BLOSUM62 substitution matrix) may be performed when too few sequences are available to produce a profile (e.g., for H3). After each inner cycle is completed, the new sequence and structure are accepted according to the Metropolis Monte Carlo criterion. After N rounds within the inner cycle, the program returns to the outer cycle, at which point the energy of the resulting design is compared to the previous design in the outer cycle. The new design is accepted or rejected according to the Monte Carlo criterion.

If optimizing the antibody-antigen orientation during the design (dock), SiteConstraints are automatically used to keep the CDRs (paratope) facing the antigen surface. These are termed ParatopeSiteConstraints. Optionally, one can enable constraints that keep the paratope of the antibody around a target epitope (antigen binding site). These are called ParatopeEpitopeSiteConstraints as the constraints are between the paratope and the epitope. The epitope is automatically determined as the interface residues around the paratope on input into the program, however, any residue(s) can be set as the epitope to limit unwanted movement and sampling of the antibody. See the examples and options below.

More detail on the algorithm can be found in the published paper.

Setup and Inputs

Antibody Design Database

This app requires the Rosetta Antibody Design Database. A database of antibodies from the original North Clustering paper is included in Rosetta and is used as the default . An updated database (which is currently updated monthly) can be downloaded here: http://dunbrack2.fccc.edu/PyIgClassify/.

It should be placed in Rosetta/main/database/sampling/antibodies/ It is recommended to use this up-to-date database.

See General-Antibody-Options-and-Tips for more. Currently, -input_ab_scheme is not supported for antibody design and an AHo-renumbered antibody must be used.


Starting Structure

The protocol begins with the three-dimensional structure of an antibody–antigen complex. Designs should start with an antibody bound to a target antigen (however optimizing just the antibody without the complex is also possible). Camelid antibodies are fully supported. This structure may be an experimental structure of an existing antibody in complex with its antigen, a predicted structure of an existing antibody docked computationally to its antigen, or even the best scoring result of low-resolution docking a large number of unrelated antibodies to a desired epitope on the structure of a target antigen as a prelude to de novo design.

The program CAN computationally design an antibody to anywhere on the target protein, but it is recommended to place the antibody at the target epitope. It is beyond the scope of this program to determine potential epitopes for binding, however servers and programs exist to predict these. Automatic SiteConstraints can be used to further limit the design to target regions.


Model Numbering

Finally, the input PDB file must be renumbered to the AHo Scheme. This can be done through the PyIgClassify Server. The PyIgClassify code can also be downloaded and run separately for local renumbering/analysis.

On input into the program, Rosetta assigns our CDR clusters using the same methodology as PyIgClassify. The RosettaAntibodyDesign protocol is then driven by a set of command-line options and a set of design instructions provided as an input file that controls which CDR(s) are designed and how. Details and example command lines and instruction files are provided below.

Command-line Examples

Basic Settings

General Design

Example 1 - GraftDesign and SeqDesign

The command-line can be as simple as:

antibody_designer.macosclangrelease -s my_ab.pdb -primary_cdrs H3 \
-graft_design_cdrs H3 -seq_design_cdrs H1 H2 -light_chain lambda -nstruct 1

This makes the H3 loop the primary CDR chosen in the outer cycle, running graft-based design on H3, while simultaneously sequence designing H1 and H2.


Example 2 - Starting with Random CDRs

Here, we want to do a denovo-run, starting with random CDRs grafted in instead of whatever we have in antibody to start with (only for the CDRs that are actually undergoing graft-design). This is useful, as we start the design with very high energy and work our way down.

antibody_designer.macosclangrelease -s my_ab.pdb -primary_cdrs H3 \
-graft_design_cdrs H3 -seq_design_cdrs H1 H2 -light_chain lambda -random_start -nstruct 1

Optimizing Interface Energy (opt-dG)

Example 1 - opt-dG

Here, we want to set the protocol to optimize the interface energy during Monte Carlo instead of total energy. The interface energy is calculated by the InterfaceAnalyzerMover through a specialized MonteCarlo called MonteCarloInterface. This is useful to improve binding energy and will result in better interface energies. Resulting models should still be pruned for high total energy. This was benchmarked in the paper, and has been used for real-life designs after - so please see it for more information.

antibody_designer.macosclangrelease -s my_ab.pdb -primary_cdrs H3 \
-graft_design_cdrs H3 -seq_design_cdrs H1 H2 -light_chain lambda -mc_optimize_dG -nstruct 1

Example 2 - Optimizing Interface Energy and Total Score (opt-dG and opt-E)

Here, we want to set the protocol to optimize the interface energy during Monte Carlo, but we want to add some total energy to the weight. Because the overall numbers of total energy will dominate the overall numbers, we only add a small weight for total energy. This has not been fully benchmarked, but if your models have very bad total energy when using opt-dG - consider using it.

antibody_designer.macosclangrelease -s my_ab.pdb -primary_cdrs H3 \
-graft_design_cdrs H3 -seq_design_cdrs H1 H2 -light_chain lambda -mc_optimize_dG \
-mc_total_weight .001 -mc_interface_weight .999 -nstruct 1

Docked Design

Example 1 - Basic DockDesign Incorporation

In this example, we use integrated RosettaDock (with sequence design during the high-res step) to sample the antibody-antigen orientation, but we don't care where the antibody binds to the antigen. Just that it binds. IE - No Constraints. The RAbD protocol always has at least Paratope SiteConstraints enabled to make sure any docking is contained to the paratope (like most good docking programs).

By default, we have now reduced the high-res dock cycles to 2/2. This should greatly speed up docking. You can change this with the -dock_first_cycles and -dock_second_cycles options. Note that for a FULL docking protocol in Rosetta, these numbers are 4 and 45 respectively.

antibody_designer.macosclangrelease -s my_ab.pdb -primary_cdrs H3 \
-graft_design_cdrs H3 -seq_design_cdrs H1 H2 -light_chain lambda -do_dock -nstruct 1

Example 2 - SiteConstraints

Allow Dock-Design, incorporating auto-generated SiteConstraints to keep the antibody around the starting interface residues. These residues are determined by being within 6A to the CDR residues.

antibody_designer.macosclangrelease -s my_ab.pdb -primary_cdrs H3 \
-graft_design_cdrs H3 -seq_design_cdrs H1 H2 -light_chain lambda -do_dock -use_epitope_constraints -nstruct 1

Example 3 - Optimizing Design to Specific Epitope Residues

Allow Dock-Design, as above, but specify the Epitope Residues and Paratope CDRs to guide design to have these interact.

antibody_designer.macosclangrelease -s my_ab.pdb -primary_cdrs H3 \
-graft_design_cdrs H3 -seq_design_cdrs H1 H2 -light_chain lambda -do_dock -use_epitope_constraints \
-paratope H3 -epitope 63A 63A:A 64 -nstruct 1

Example 4 - DeNovo Design

Here, we want to do a denovo-run, creating an interface at the light-chain, starting with random CDRs grafted in instead of whatever we have in the antibody to start with (for the designing CDRs).

antibody_designer.macosclangrelease -s my_ab.pdb -primary_cdrs L1 L2 L3 \
-graft_design_cdrs L1 L2 L3 -seq_design_cdrs L1 L2 L3 -light_chain lambda -do_dock \
-use_epitope_constraints -paratope L1 L2 L3 -epitope 63A 63A:A 64 -random_start -dock_first_cycles 1 -dock_second_cycles 2

Instruction File Customization

More complicated design runs can be created by using the Antibody Design Instruction file. This file allows complete customization of the design run. See below for a review of the syntax of the file and possible customization.

antibody_designer.macosclangrelease -s my_ab.pdb -primary_cdrs H3 \
-graft_design_cdrs H3 -seq_design_cdrs H1 H2 -light_chain lambda \
-cdr_instructions my_instruction_file.txt -nstruct 1

Advanced Settings

Example 1 - Incorporating a Resfile to Limit Design

Here, we will disallow ANY sequence design into Proline residues and Cysteine residues, while giving a resfile to further LIMIT design and packing as specific positions. These can be given as 3 or 1 letter codes and mixed codes such as PRO and C are accepted. Note that the resfile does NOT turn any residues ON, it is simply used to optionally LIMIT design residue types and design and packing positions.

antibody_designer.macosclangrelease -s my_ab.pdb -primary_cdrs H3 \
-graft_design_cdrs H3 -seq_design_cdrs H1 H2 -light_chain lambda \
-resfile my_resfile.resfile -dissallow_aa PRO CYS -nstruct 1

Example 2 - Flexible Backbone Design (Generally Recommended)

Here, we will change the mintype to relax. This mintype enables Flexible-Backbone design. Our default is to use min/pack cycles, but relax typically works better. However, it also takes considerably more time!

antibody_designer.macosclangrelease -s my_ab.pdb -primary_cdrs H3 \
-graft_design_cdrs H3 -seq_design_cdrs H1 H2 -light_chain lambda \
-resfile my_resfile.resfile -dissallow_aa PRO CYS -mintype relax -nstruct 1

Example 3 - Incorporating Framework Design

Finally, we want to allow the framework residues AROUND the CDRs we will be designing and any interacting antigen residues to design as well here. We will disable conservative framework design as we want something funky (this is not typically recommended and is used here to indicate what you CAN do.

antibody_designer.macosclangrelease -s my_ab.pdb -primary_cdrs H3 \
-graft_design_cdrs H3 -seq_design_cdrs H1 H2 -light_chain lambda \
-resfile my_resfile.resfile -dissallow_aa PRO CYS -mintype relax \
-design_antigen -design_framework -conservative_framework_design false -nstruct 1

Expert Settings

Example 1 - H3 Design

Now, we will spice things up even further. We are feeling daring today. A new Rosetta energy function with fully polarizable forcefields has just been published, we have our first quantum computer, Andrew just got done the Quantum JD through JD4, and we have LOTS of money for designs (I can dream, right! ;). We are ready to put Rosetta to the test.

We will enable H3 Stem design here, which can cause a flipping of the H3 stem type from bulged to non-bulged and vice-versa. Typically, if you do this, you may want to run loop modeling on the top designs to confirm the H3 structure remains in-tact.

antibody_designer.macosclangrelease -s my_ab.pdb -primary_cdrs H3 \
-graft_design_cdrs H3 -seq_design_cdrs H1 H2 -light_chain lambda -design_H3_stem -nstruct 1

Cool. That should make some interesting antibodies for our experiment.


Example 3 - Increasing MonteCarlo Temperature

Now, we will change around the KT to get more interesting samplings (from our 1.0 default).

antibody_designer.macosclangrelease -s my_ab.pdb -primary_cdrs H3 \
-graft_design_cdrs H3 -seq_design_cdrs H1 H2 -light_chain lambda \
-design_H3_stem -inner_KT 2.0 -outer_KT 2.0 -nstruct 1

Example 4 - Increasing Sequence Design Variability

Finally, we want increased variability for our sequence designs. So, we will increase number of sampling rounds for our lovely cluster profiles using the -seq_design_profile_samples option.

Description of the option (default 1): "If designing using profiles, this is the number of times the profile is sampled each time packing done. Increase this number to increase variability of designs - especially if not using relax as the mintype."

antibody_designer.macosclangrelease -s my_ab.pdb -primary_cdrs H3 \
-graft_design_cdrs H3 -seq_design_cdrs H1 H2 -light_chain lambda \
-design_H3_stem -inner_KT 2.0 -outer_KT 2.0 -seq_design_profile_samples 5 -nstruct 1

Antibody Design CDR Instruction File

The Antibody Design Instruction File handles CDR-level control of the algorithm and design. It is used to create the CDRSet for sampling whole CDRs from the PDB, as well as fine-tuning the minimization steps and sequence design strategies.

For example, in a redesign project, we may only want to design a particular CDR and explore the local conformations within its starting CDR cluster, or we may simply want to optimize the sequence of a CDR or set of CDRs using our cluster profiles. These examples can be accomplished using the CDR Instruction File. This file uses a simple syntax where each CDR is controlled individually in the first column, and then a rudimentary language controls each part of the antibody design machinery. The file controls which CDRs are sampled, what the final CDRSet will be, whether sequence design is performed and how, and which minimization type will be used to optimize the structure and sequence for each CDR. Some of these commands can also be controlled via command-line options for simple-to-setup runs. Any option set via command-line (such as -graft_design_cdrs and -seq_design_cdrs will override anything set in the instruction file)

Syntax

The CDR Instruction file is composed of tab or white-space delimited columns. The first column on each line specifies which CDR the option is for. For each option, 'ALL' can be given to control all of the CDRs at once. ALL can also be used in succession to first set all CDRs to something and then one or more CDRs can be set to something else in succeeding lines. Commands are not lower- or upper-case-sensitive. A ‘#’ character at the beginning of a line is a file comment and is skipped.

First column - CDR,

Second Column - TYPE

Other columns can be used to specify lists, etc.

Instruction Types

Type Description
GraftDesign Instructions for the Graft-based Design stage
SeqDesign Instructions for Sequence Design
CDRSet Instructions for which structures end up in the set of structures from which to sample during GraftDesign. option follows the syntax: CDR CDRSet option
MinProtocol Instructions for Minimization

Example

#H3 no design; just flexible motion of the loop

#SET All to design first
ALL GraftDesign ALLOW
ALL SeqDesign ALLOW

#Fix H3 and disallow GraftDesign for H1 and L2
L2 GraftDesign FIX
H1 GraftDesign FIX

#Disallow GraftDesign and SeqDesign for H3
H3 FIX

ALL MinProtocol MINTYPE relax

ALL CDRSet EXCLUDE PDBIDs 1N8Z 3BEI

#ALL CDRSet EXCLUDE Clusters L1-11-1 L3-9-cis7-1 H2-10-1

General Design Syntax

Instruction Description
L1 Allow Allow L1 to Design in both modes
L1 GraftDesign Allow Allow L1 to GraftDesign
L1 SeqDesign Allow Allow L1 to SequenceDesign
L1 Fix Disable L1 to Design in both modes
L1 GraftDesign Fix Disable L1 to GraftDesign
L1 SeqDesign Fix Disable L1 to SequenceDesign
L1 Weights Z Weight a particular CDR when choosing which CDR to design. By default, all CDRs have an equal weight.

CDRSet Syntax (General)

Instruction Description
L1 CDRSet Length Max X Maximum length of CDR
L1 CDRSet Length Min Y Minimum length of CDR
L1 CDRSet Cluster_Cutoffs Z Limits the CDRSet to include only CDRs of clusters that have Z or more members. Used in conjunction with Sequence Design cutoffs to sample clusters where profile data is sufficient or simply common clusters of a particular species.
L1 CDRSet ONLY_CURRENT_CLUSTER Limits the CDRSet to include only CDRs belonging to the identified cluster of the starting CDR. Overrides other options. Used for sequence and structure sampling within CDR clusters. Useful for redesign applications.
L1 CDRSet Center_Clusters_Only Limits the CDRSet to include only CDRs that are cluster centers. Overrides other options. Used for broadly sampling CDR structure space. Useful to get an idea of what CDRs can fit well or for a first-pass run for de novo design.

CDRSet Syntax (Include and Excludes)

This set of options either includes only the items specified or it excludes specific items. The syntax is thus:

L1 CDRSet INCLUDE_ONLY option list of items

OR

L1 CDRSet EXCLUDE option list of items

The options for this type of CDRSet syntax are listed below

Option Description
PDBIDs Include only or leave out a specific set of PDBIds. This is very useful for benchmarking purposes.
Clusters Include only or leave out a specific set of Clusters. Useful if the user already knows which clusters are able to interact with antigen, whether this is from previous runs of the program or via homologues. This is also useful for benchmarking.
Germline Include only or leave out specific human/mouse germlines.
Species Include only or leave out specific species. Very useful limiting possible immune reactions in the final designs. 2 Letter designation as used by IMGT. Hu and Mo for Human and Mouse.

Sequence Design Instructions

The Sequence Design Instructions, used with the keyword SeqDesign after the CDR specifier, is used to control the sequence design aspect of the algorithm. In addition to controlling which CDRs undergo sequence design, the Sequence Design options control which strategy to use when doing sequence design (primary strategy), and which strategy to use if the primary strategy cannot be used for that CDR (fallback strategy).

Currently, the fallback strategy is used if the primary strategy does not meet statistical requirements. For example, using cluster-based profiles for sequence design (which is the default), it is imperative that the particular cluster has at least some number of sequences in the database for the strategy to be successful. By default, this cutoff is set at 10 and uses the command-line option, ‑seq_design_stats_cutoff Z. This means that if a cluster has less than 10 members and the primary strategy is to use cluster-based profiles, then the fallback strategy will be used. The default fallback strategy for all of the CDRs is the use of a set of conservative mutations with data coming from the BLOSUM62 matrix by default, with the set of mutations allowed consisting of those substitutions with positive or zero scores in the matrix. For example, if we have a D at some position for which we are using this fallback strategy, the set of allowed mutants would be N, Q, E, and S. Further, the set of conservative mutations can be controlled using the command-line option, ‑cons_design_data_source source, where other BLOSUM matrices can be specified. All BLOSUM matrices can be used for conservative design, where the numbers (40 vs. 62 vs. 80 etc.) indicate the sequence similarity cutoffs used to derive the BLOSUM matrices - with higher numbers being a more conservative set of mutations.

General Syntax

Option Description
L1 SeqDesign Strategy Z Set Primary Sequence Design Strategy
L1 SeqDesign Fallback_Strategy Z Set Fallback Sequence Design Strategy

Strategy Types

Z Strategy Use Description
Profiles Primary Use cluster-based profiles based on sequence probabilities as described above
Profile_Sets Primary Randomly sample a full CDR sequence from the CDR cluster each time packing is done
Profiles_Combined Primary Randomly sample a full CDR sequence from the CDR cluster and sample from the cluster-based profiles each time packing is done. Used to increase variability. Helpful in conjunction with center cluster member CDRSet sampling
Conservative Both Conservative design operation as described above
Basic Both Basic design – no profiles, just enabling all amino acids at those positions
Disable Fallback Turn off design for that CDR if primary strategy does not meet cutoffs.

Etc

Option Description
L1 SeqDesign DISALLOWED Disable specific amino acid sampling for this CDR (One or Three letter codes. Can be mixed) Ex. ARG C S ASN PRO

Minimization

Many minimization types are implemented; however, they each require a different amount of time to run. See the MinType descriptions below for a list of acceptable options and their use.

Option Description
L1 MinProtocol MinType Z Set the minimization type for this CDR
L1 MinProtocol MinOther CDRX, CDRY, CDRZ Set other CDRs to minimize during the optimization/design of this particular CDR. A packing shell around the CDRs to be minimized within a set distance (default 6 Å) is created. Any CDRs or regions (framework/antigen) set to design within this shell will be designed. These regions will use all set sequence design options (profiles, conservative, etc.). This option is useful for creating CDR-CDR interactions for loop and antibody structural stability. By default, we reasonably minimize other CDRs during the optimization step. (More CDRs = Lower Speed)

Default Settings

These settings are overridden by your set instruction file/command-line options


#General
ALL FIX
DE FIX

ALL WEIGHTS 1.0

#Minimization - Options: min, relax, ds_relax cartesian, backrub, repack, none
ALL MinProtocol MINTYPE min

#Neighbors.  These are conservative values used in benchmarking.  Limit these to speed up runs.
L1 MinProtocol Min_Neighbors L2 L3
L2 MinProtocol Min_Neighbors L1
L3 MinProtocol Min_Neighbors L1 H3
H1 MinProtocol Min_Neighbors H2 H3
H2 MinProtocol Min_Neighbors H1
H3 MinProtocol Min_Neighbors L1 L3

#Length
ALL CDRSet LENGTH MAX 25
ALL CDRSet LENGTH MIN 1


#Profile Design
ALL SeqDesign STRATEGY PROFILES
ALL SeqDesign FALLBACK_STRATEGY CONSERVATIVE

##DE Loops can be designed.  They use conservative design by default since we have no profile data!
DE SeqDesign STRATEGY CONSERVATIVE

GraftDesign Sampling Algorithms

These change the way CDRs are sampled from the antibody design database. They can be specified using the -design_protocol flag.

-design_protocol Description
even_cluster_mc Evenly sample clusters during the GraftDesign stage by first choosing a cluster from all the clusters set to design for the chosen Primary CDR and then choosing a structure within that cluster. (DEFAULT)
even_length_cluster_mc Evenly sample lengths and clusters during the GraftDesign stage by first choosing a length from the set of lengths for the chosen Primary CDR and the a cluster from the set of clusters, and then finally a structure within that cluster. Useful to broaden set of lengths sampled during the protocol.
gen_mc Sample CDRs to GraftDesign according to their distribution in the database. This results in common clusters and lengths being sampled more frequently. However, these lengths/clusters may not be those regularly seen in nature vs regularly crystalized. AKA - they are biased towards crystals, however, they have more profile data associated with them.
deterministic_graft Deterministic Graft is meant to try every CDR combination from the CDRSet (the set of clusters and structures). The outer loop is done deterministically for each CDR in a set. It is very useful for trying small numbers of combinations - such as all loop lengths >=12 for H2 or all CDRs of a particular cluster. Note that there is no outer monte carlo, so the final designs are the best found by the protocol, and each sampling is independent from the others. If you have too many structures in your CDRSet (such as all L1) and you try combos that are beyond a certain limit (AKA - they will never finish), you will error out. Once a Multi-Threaded Rosetta is working (should be early 2018), trying all possibilities is certainly something that is more plausible. If you are interested in using something like this, please email the author.

Structure Optimization Types

These 'Mintypes' can be independently set for each CDR through the instruction file or generally set using the command line option -mintype. The default is min, as this has some optimization and does not take a very long time (and has been shown to be comparable to relax). Although we refer to 'design' we mean side-chain packing, with any residues/CDRs set to design as designing. For further information on the algorithm and strategies used for sequence design, please see the instruction file overview and the methods section of the paper.

Circular Harmonic Dihedral Constraints are added to each CDR according the cluster of the CDR or the starting dihedrals if this is a rare cluster that has no data. These ensure that minimization and design does not destroy the loop.

-mintype Description
min Cycle of design->min->design->min. Results in good structures, however not as good as relax in recovering native physical characteristics. Significantly faster. (DEFAULT)
cartmin Cycle of design->min->design->min Cartesian Space Minimization. Automatically adds cart_bonded term if not present and turns off pro_close
relax Flexible Backbone design using RelaxedDesign, which is neighbor-aware design during FastRelax where the packing shell to the designing CDRs is updated at every packing iteration. Results in lower energies and closer physical characteristics to native, but takes significantly longer. It is recommended to first run min and then relax mintype on the top resulting models. (Relax Protocol Paper)
dualspace_relax Flexible Backbone design using 'RelaxedDesign' while optimizing both Dihedral and Cartesian space Dualspace Relax Protocol Paper
backrub Cycle of backrub->design. Uses backrub to optimize the Backbone of the CDRs set to minimize. Use the -add_backrub_pivots 11A 12A 12A:B option to add additional sets of back rub pivots, such as to add flexibility to the antigen interface. Flexibility is extremely minimal, but in some cases may be useful. (Original Backrub Protocol Paper)

RosettaScripts and PyRosetta

The Full protocol that is the application is available to RosettaScripts as the AntibodyDesignProtocol. This just has a few extra options before and after design such as running fast relax and or snug dock.

The Configurable main Mover is available as the AntibodyDesignMover.

Individual components of RAbD can be used to create your own custom antibody modeling and design protocols in RosettaScripts (or PyRosetta).

Command-line Options

General Options

Option Description
-view Enable viewing of the antibody design run. Must have built using extras=graphics and run with antibody_designer.graphics executable (Default=False)
-cdr_instructions Path to CDR Instruction File
-antibody_database _Path to the current Antibody Database, updated weekly. Download from http://dunbrack2.fccc.edu/PyIgClassify/ _ (Default=/sampling/antibodies/antibody_database_rosetta_design.db)
-paper_ab_db Force the use the Antibody Database with data from the North clustering paper. This is included in Rosetta. If a newer antibody database is not found, we will use this. The full ab db is available at Through PyIgClassify (Default = false)

Basic settings for an easy-to-setup run

Option Description
-seq_design_cdrs Enable these CDRs for Sequence-Design. (Can be set here or in the instructions file. Overrides any set in instructions file if given ) Ex -seq_design_cdrs L1 L2 L3 h1
-graft_design_cdrs Enable these CDRs for Graft-Design. (Can be set here or in the instructions file. Overrides any set in instructions file if given) Ex -graft_design_cdrs L1 L2 L3 h1
-primary_cdrs Manually set the CDRs which can be chosen in the outer cycle. Normally, we pick any that are sequence-designing.
-mintype The default mintype for all CDRs. Individual CDRs may be set via the instructions file (Options = min, cartmin, relax, backrub, pack, dualspace_relax, cen_relax, none) (Default=min)
-disallow_aa Disallow certain amino acids while sequence-designing (could still be in the graft-designed sequence, however). Useful for optimizing without, for example, cysteines and prolines. Applies to all sequence design profiles and residues from any region (cdrs, framework, antigen). You can control this per-cdr (or only for the CDRs) through the CDR-instruction file. A resfile is also accepted if you wish to limit specific positions directly. (Three or One letter codes)
-top_designs Number of top designs to keep (ensemble). These will be written to a PDB and each move onto the next step in the protocol. (Default=1)
-do_dock Run a short lowres + highres docking step in the inner cycles. (dock/min). Recommended 2 inner cycles for better coverage. (dock/min/dock/min). Inner/Outer loops for highres are hard coded, while low-res can be changed through regular low_res options. If sequence design is enabled, will design regions/CDRs set during the high-res dock (Default=false)

Energy Optimization settings (opt-dG)

Option Description
-mc_optimize_dG Optimize the dG during MonteCarlo. It is not possible to do this within overall scoring, but where possible, do this during MC calls. This option does not globally-use the MonteCarloInterface object, but is protocol-specific. This is due to needing to know the interface it will be used on. dG is measured by the InterfaceAnalyzerMover. (Default=false)
-mc_interface_weight Weight of interface score if using MonteCarloInterface with a particular protocol. (Default=1.0)
-mc_total_weight Weight of total score if using MonteCarloInterface with a particular protocol. (Default=0.0)

Protocol Rounds

Option Description
-outer_cycle_rounds Rounds for outer loop of the protocol (not for deterministic_graft ). Each round chooses a CDR and designs. One run of 100 cycles with relax takes about 12 hours. If you decrease this number, you will decrease your run time significantly, but your final decoys will be higher energy. Make sure to increase the total number of output structures (nstruct) if you use lower than this number. Typically about 500 - 1000 nstruct is more than sufficient. Full DeNovo design will require significantly more rounds and nstruct. If you are docking, runs take about 30 percent longer. (Default=25)
-inner_cycle_rounds Number of times to run the inner minimization protocol after each graft. Higher (2-3) rounds recommended for pack/min/backrub mintypes or if including dock in the protocol. (Default = 1)
-dock_cycle_rounds Number of rounds for any docking. If you are seeing badly docked structures, increase this value. (Default=1)
-dock_first_cycles Number of first cycles for High-Res docking step. (Default=2)
-dock_second_cycles Number of second cycles for High-Res docking step. (Default=2)

Distance Detection

Option Description
-interface_dis Interface distance cutoff. Used for repacking of interface, epitope detection, etc. (Default=8.0)
-neighbor_dis Neighbor distance cutoff. Used for repacking after graft, minimization, etc. (Default=6.0)

Paratope + Epitope

Option Description
-paratope Use these CDRs as the paratope. Default is all of them. Currently only used for SiteConstraints. Note that these site constraints are only scored docking unless -enable_full_protocol_atom_pair_cst is set (Ex -paratope L1 h1)
-epitope Use these residues as the antigen epitope. Default is to auto-identify them within the set interface distance at protocol start if epitope constraints are enabled. Currently only used for constraints. PDB Numbering. Optional insertion code. Example: 1A 1B 1B:A. Note that these site constraints are only used during docking unless -enable_full_protocol_atom_pair_cst is set.
-use_epitope_constraints Enable use of epitope constraints to add SiteConstraints between the epitope and paratope. Note that paratope constraints are always used. Note that these site constraints are only used during docking unless -global_atom_pair_cst_scoring is set.(Default = false)

Regional Sequence Design

Option Description
-design_antigen Design antigen residues during sequence design. Intelligently. Typically, only the neighbor antigen residues of designing cdrs or interfaces will be co-designed. Useful for specific applications.(Default = false)
-design_framework Design framework residues during sequence design. Typically done with only neighbor residues of designing CDRs or during interface minimization. (Default = false)
-conservative_framework_design 'If designing Framework positions, use conservative mutations instead of all of them.'(Default=true)

Seq Design Control

Option Description
-resfile Use a resfile to further limit which residues are packed/designed, and can further limit residue types for design.
-design_H3_stem Enable design of the first 2 and last 3 residues of the H3 loop if sequence designing H3. These residues play a role in the extended vs kinked H3 conformation. Designing these residues may negatively effect the overall H3 structure by potentially switching a kinked loop to an extended and vice versa. Rosetta may get it right. But it is off by default to err on the cautious side of design. Sequence designing H3 may be already risky. (Default=false)
-design_proline Enable proline design. Profiles for proline are very good, but designing them is a bit risky. Enable this if you are feeling daring. (Default=false)
-sample_zero_probs_at Value for probabilstic design. Probability that a normally zero prob will be chosen as a potential residue each time packer task is called. Increase to increase variablility of positions. (Default=0)

Profile Stats

Option Description
-seq_design_stats_cutoff Value for probabilistic -> conservative sequence design switch. If number of total sequences used for probabilistic design for a particular cdr cluster being designed is less than this value, conservative design will occur. More data = better predictability. (Default=10)
-seq_design_profile_samples If designing using profiles, this is the number of times the profile is sampled each time packing done. Increase this number to increase variability of designs - especially if not using relax as the mintype. (Default=1)

Constraint Control

Option Description
-dihedral_cst_weight Weight to use for CDR CircularHarmonic cluster-based or general constraints that are automatically added to each structure and updated after each graft. Set to zero if you dont want to use these constraints. Note that they are not used for the backrub mintype. Overrides weight/patch settings.(Default = .3)
-atom_pair_cst_weight Weight to use for Epitope/Paratope SiteConstraints. Paratope Side contraints are always used. Set to zero to completely abbrogate these constraints. Overrides weight/patch settings.'Real' (Default = 0.01)
-global_dihedral_cst_scoring Use the dihedral cst score throughout the protocol, including final scoring of the poses instead of just during minimization step (Default = false)
-global_atom_pair_cst_scoring Use the atom pair cst score throughout the protocol, including final scoring of the poses instead of just during docking. Typically, the scoreterm is set to zero for scorefxns other than docking to decrease bias via loop lengths, relax, etc. It may indeed help to target a particular epitope quicker during monte carlo design if epitope constraints are in use, as well for filtering final models on score towards a particular epitope if docking. (Default = true)

Fine Control

Option Description
-idealize_graft_cdrs Idealize the CDR before grafting. May help or hinder. (Default = false)
-add_backrub_pivots 'Additional backrub pivot residues if running backrub as the MinType. PDB Numbering. Optional insertion code. Example: 1A 1B 1B:A. Can also specify ranges: 1A-10:A. Note no spaces in the range.
-inner_kt KT used in the inner min monte carlo after each graft. (default = 1.0),
-outer_kt KT used for the outer graft Monte Carlo. Each graft step will use this value (Default = 1.0),

Outlier Control

Option Description
-use_outliers Include outlier data for GraftDesign, profile-based sequence design stats, and cluster-based dihedral constraints. Outliers are defined as having a dihedral distance of > 40 degrees and an RMSD of >1.5 A to the cluster center. Use to increase sampling of small or rare clusters. (Default=false)
-use_H3_graft_outliers Include outliers when grafting H3. H3 does not cluster well, so most structures have high dihedral distance and RMSD to the cluster center. Due to this, cluster-based dihedral constraints for H3 are not used. Sequence profiles can be used for clusters, but not usually. (Default = true)
-use_only_H3_kinked Remove any non-kinked CDRs from the CDRSet if grafting H3. For now, the match is based on the ramachandran area of the last two residues of the H3. Kinked in this case is defined as having AB or DB regions at the end. Will be improved for detection (Default = false)

Protocol Steps

Option Description
-design_protocol Set the main protocol to use. Note that deterministic is currently only available for the grafting of one CDR. (Options = gen_mc, even_cluster_mc, even_length_cluster_mc, deterministic_graft)(Default=even_cluster_mc)
-run_interface_analyzer Run the Interface Analyzer and add the information to the resulting score function for each top design output. (Default = true)

Memory management / CDRSet caching

Option Description
-high_mem_mode If false, we load the CDRSet (CDRs loaded from the database that could be grafted) on-the-fly for a CDR if it has more than 50 graft-design members. If true, then we cache the CDRSet before the start of the protocol. Typically, this will really only to come into play when designing all CDRs. For de-novo design of 5/6 CDRs, without limiting the CDRSet in the instructions file, you will need 3-4 gb per process for this option. (default = false)
-cdr_set_cache_limit If high_mem_mode is false, this is the limit of CDRSet cacheing we do before we begin load them on-the-fly instead. If high_mem_mode is true, then we ignore this setting. If you have extremely low memory per-process, lower this number (default = 300)

Benchmarking

Option Description
-random_start Start graft design (currently) with a new set of CDRs from the CDRSets as to not bias the run with native CDRs. (Default=false)
-remove_antigen Remove the antigen from the pose before doing any design on it (Default = false)
add_graft_log_to_pdb Add the full graft log to the output pose. Must also pass -pdb_comments option. (Default = 'true')

See Also