Talks and presentations

Next-generation antibody modeling

February 15, 2015

Invited Seminar, Institute for Cellular and Molecular Biology, University of Texas at Austin, Austin, TX

Developments in high-throughput sequencing technologies have made it possible to sequence 105 B cells in a single experiment, renewing the need for accurate structural modeling methodologies. I discuss the process by which we identify weaknesses in current approaches and how we go about filling in the gaps.

Next-generation antibody modeling

July 29, 2014

Invited Seminar, Center for Biomolecular Structure and Dynamics, University of Montana, Missoula, MT

Developments in high-throughput sequencing technologies have made it possible to sequence 105 B cells in a single experiment, renewing the need for accurate structural modeling methodologies. I discuss the process by which we identify weaknesses in current approaches and how we go about filling in the gaps.

The origin of CDR H3 structural diversity

May 06, 2014

Vortrag, Fakultät für Chemie, Technische Universität München, Munich, Germany

The third complementarity determing region loop on the heavy chain (CDR H3) of antibodies is the most diverse of the CDR loops in terms of length, sequence and structure. This loop also provides a plurality of interatomic contacts and binding energy in antibody–antigen complexes. In light of this, I disucss the factors that contribute to the observed structural diversity of CDR H3 in order to provide a starting point for improved strucutre prediction methods.

Computational structure prediction, docking and design of antibodies

December 11, 2013

Invited Talk, IBC Antibody Engineering and Therapeutics, Huntington Beach, CA

I discussed the latest developments in Rosetta-based approaches for antibody structure prediction and docking, with a focus on new loop prediction methods, (specifically for CDR H3 loops), and using homology models to design for binding desired target epitopes.

Benchmarking RosettaAntibody: Antibody Modeling Assessment II

December 08, 2013

Workshop, IBC Antibody Engineering and Therapeutics, Huntington Beach, CA

Antibody Modeling Assessment II (AMA-II) presented an opportunity to test the RosettaAntibody structure prediction tool on 11 benchmark antibody FV regions whose structures were determined, but not yet deposited in the Protein Data Bank. Along with groups from Accelrys Software, Inc, Chemical Computing Group, Inc, Astellas Pharma, Macromoltek, and Schrödinger, we discussed the strengths, weaknesses, and future plans for each of the methods employed.

Kinked CDR H3-like loops are common

November 06, 2013

Conference Talk, AIChE Annual Conference, San Francisco, CA

The difficulty of de novo CDR H3 loop modeling is surprising in many cases because of the modest loop lengths at which they occur. One possible explanation is that V(D)J recombination can produce loops that access conformations that are extremely rare in existing protein structural databases. An alternate hypothesis is that the environment formed by the VH and VL domains stabilizes CDR H3 loop conformations that existing methods do not detect as favorable. We identified a diverse set of loops across a wide range of lengths that adopt H3-like conformations. These loops show that the kinked conformation of CDR H3 loops is common.

Antibodies are proteins too!

July 28, 2013

Conference Talk, Rosetta Conference, Leavenworth, WA

The third complementarity-determining region loop on the IgG heavy chain (CDR H3) is the focal point of the gene recombination that gives rise to the immense diversity of antibodies. A consequence of this diversity is observed in the variability in length, sequence and structure of the H3 loop. Because of the circumstances that give rise to the H3 loop, it is unclear if the conformations they adopt are unique to antibodies. We set out to determine what it means for a loop to have an H3-like conformation and whether or not these conformations are commonplace in other proteins. To that end, we searched a set of high-quality non-antibody structures for regions with 1) 3D transformations matching the conserved transformation of the H3 loop and 2) a C-terminal “kink”/“bulge”. We found these criteria were enough to identify a set of loops with similar DSSP code distributions and RMSDs within 2.0 Å of nearly all H3 loops of length 15 or shorter. We believe this set can be used to find starting structures for H3 loop modeling and that this result is powerful evidence suggesting that CDR H3 is just another loop.

PyRosetta 2.0: I can make a new score term in 6 lines!

August 06, 2011

Conference Talk, Rosetta Conference, Leavenworth, WA

Along with Sergey Lyskov, I discuss the new developments in the python-accessible distribution of Rosetta, PyRosetta. Of particular note is that every component of Rosetta is now available in PyRosetta and key Rosetta data structures can be created in python and passed back into the system, enabling rapid prototyping of new score function terms and sampling strategies.

Using PyRosetta for research

August 04, 2010

Workshop, Rosetta Conference, Leavenworth, WA

Andrew Leaver-Fay, Dan Kulp, and Sergey Lyskov, and I led a workshop that went over first time use of PyRosetta, using interactive sessions to get things done quickly, and basic usage patterns.

PyMOL–PyRosetta Integration

August 04, 2010

Conference Talk, Rosetta Conference, Leavenworth, WA

In this talk I present a new tool to enable real-time visualization in PyMOL of macromolecular simulations performed in Rosetta or PyRosetta.