Leucine Radical

The leucine radical (C₄H₉), also known as its side chain, is a nonpolar, aliphatic hydrocarbon structure that distinguishes leucine from other amino acids. It consists of a branched isobutyl group, making it hydrophobic and strongly involved in the stabilization of protein cores through hydrophobic interactions. This radical does not participate in hydrogen bonding or ionic interactions, but its branched structure contributes significantly to the tertiary structure of proteins. Along with the amino (–NH₂) and carboxyl (–COOH) groups, the C₄H₉ radical completes leucine’s structure, giving it an important role in protein folding, membrane association, and energy metabolism.

Description of the dataset

This dataset contains various quantum properties that represent and describe the leucine radical under certain conditions. The dataset includes molecular geometries, the Hamiltonian, energies, and other descriptors that can be used to analyze the molecule’s behavior and build the entire leucine amino acid.

Key features include:

• Molecular information (name, molecular formula, charge, spin, geometry, etc.)
• Resource estimation (number of atoms, electrons, orbitals, qubits and Hamiltonian coefficients to represent the molecule)
• Jordan-Wigner Hamiltonian representation
• Approximations to the ground state energy

This dataset is intended to:

• Facilitate research on relevant organic molecules by providing crucial yet computationally expensive properties—such as ground state energy and the molecular Hamiltonian—thereby accelerating advancements in quantum simulations of biomolecules.

• Enhance the characterization of larger biomolecular systems by bridging the gap between existing datasets, which are primarily focused on small molecules, and the needs of researchers working on peptides and proteins.

• Support hybrid QM/ML approaches, allowing researchers to train models that accurately and efficiently predict the properties of larger and more complex systems.

• Accelerate drug discovery and biomolecular research, as proteins are central to numerous biological and therapeutic processes.

• Enable the study of fragmentation and reassembly techniques by proposing new chemical corrections for bond formation and ensuring accurate reconstruction of molecular properties after simulation, in line with the results obtained in our latest work.

Example Usage

To save space, the Hamiltonian in this dataset is broken up into parts and stored as a string. In the following example, we reconstruct a PennyLane Hamiltonian from these parts.

import pennylane as qml
coefficients = []
operators = []
hamiltonian_chunks = []

# Download the dataset and retrieve the Hamiltonian chunks
ds = qml.data.load('other', name='r-leu')
for key in ds.list_attributes():  # Sort the keys to preserve the correct sequence
    if "hamiltonian" in key:
        hamiltonian_chunks.append(getattr(ds, key))  # Decode bytes to string

# Combine all Hamiltonian chunks into a single string
full_hamiltonian = "".join(hamiltonian_chunks)

# Helper function to convert a string representation into a PennyLane operator
def string_to_operator(op_string):
    if "Identity" in op_string:
        return qml.Identity(0)  # Identity defaults to acting on qubit 0
    
    terms = op_string.split(" @ ")  # Separate tensor product terms
    ops = []
    
    for term in terms:
        try:
            op, wire = term.split("(")
            wire = int(wire.strip(")"))  # Extract the qubit index
            if op == "X":
                ops.append(qml.PauliX(wire))
            elif op == "Y":
                ops.append(qml.PauliY(wire))
            elif op == "Z":
                ops.append(qml.PauliZ(wire))
        except ValueError:
            continue  # Skip malformed lines
    
    return qml.prod(*ops) if len(ops) > 1 else ops[0]  # Create composite operator if needed

# Process each line of the combined Hamiltonian string
for line in full_hamiltonian.split("\n"):
    line = line.strip()
    if not line or "Coefficient" in line or "Operators" in line:
        continue  # Skip empty lines or header lines

    parts = line.split()
    
    try:
        coeff = float(parts[0])  # Extract the coefficient
        op_string = " ".join(parts[1:])  # Extract the operators
        coefficients.append(coeff)
        operators.append(string_to_operator(op_string))
    except ValueError:
        continue  # Gracefully handle conversion errors

# Build the PennyLane Hamiltonian
hamiltonian = qml.Hamiltonian(coefficients, operators)

version 0.1 : initial public release