Dear all,
Systems Biology lab group meeting will take place on
Monday, October 4, 2021
10:00am ET
VIRTUAL
Presenter:
Vikram Mulligan, Ph.D., Research Scientist, Systems Biology Group,
Center for Computational Biology, Flatiron Institute
Advancements in quantum computational approaches for biomolecular design
Biological heteropolymers, including glycans, nucleic acids, and proteins, are difficult to model computationally, in part because such modeling frequently involves hard search problems in vast, multidimensional solution spaces. If one wishes to design an N-amino acid peptide with a particular fold and function, for example, and one has D amino acid building blocks from which to choose, there are D to the power of N possible sequences to search for an optimal sequence. On conventional computers, we use heuristic methods such as simulated annealing-based searches to make these problems tractable, but the exponentially-scaling search problem prevents application of these methods to problems with large N or large D (or both). We previously showed that a viable alternative is to solve these problems using an adiabatic quantum annealer, a type of special-purpose quantum computer. Using this approach, we have designed and, with collaborators, synthesized folding peptides. In this talk, I will present an updated approach for solving these problems on the quantum annealer, one that uses far fewer quantum bits (qubits) and permits challenging design problems to be solved on current hardware. I will also present early work with a collaborator to map these problems to general-purpose gate-based quantum computers.
Systems Biology lab group meeting will take place on
Monday, October 4, 2021
10:00am ET
VIRTUAL
Presenter:
Vikram Mulligan, Ph.D., Research Scientist, Systems Biology Group,
Center for Computational Biology, Flatiron Institute
Advancements in quantum computational approaches for biomolecular design
Biological heteropolymers, including glycans, nucleic acids, and proteins, are difficult to model computationally, in part because such modeling frequently involves hard search problems in vast, multidimensional solution spaces. If one wishes to design an N-amino acid peptide with a particular fold and function, for example, and one has D amino acid building blocks from which to choose, there are D to the power of N possible sequences to search for an optimal sequence. On conventional computers, we use heuristic methods such as simulated annealing-based searches to make these problems tractable, but the exponentially-scaling search problem prevents application of these methods to problems with large N or large D (or both). We previously showed that a viable alternative is to solve these problems using an adiabatic quantum annealer, a type of special-purpose quantum computer. Using this approach, we have designed and, with collaborators, synthesized folding peptides. In this talk, I will present an updated approach for solving these problems on the quantum annealer, one that uses far fewer quantum bits (qubits) and permits challenging design problems to be solved on current hardware. I will also present early work with a collaborator to map these problems to general-purpose gate-based quantum computers.
