2
A. Monari and X. Assfeld
is also of seminal importance in the growing field of phototherapy [5]. For instance
small organic or organometallic drugs can interact with DNA and induce permanent
lesions of the nucleic acid once activated by exposition to light [6–8], so as to
constitute very efficient photo-chemo-therapeutic agents. Taking care of the environment in theoretical calculations, both for ground and excited states is thus mandatory if one wants to obtain realistic results, i.e. directly comparable to experimental data, or to make reliable predictions [9–13].
The surrounding, i.e. the solvent or the bio-macromolecule, is a very large system, containing at least thousands of atoms to obtain an acceptable representation.
Even if the progresses of quantum chemistry over the past decades are tremendous (algorithms, computer, new method, …), describing such large systems with
quantum methods is still far beyond our computational capacities, especially when
dealing with electronically excited states. We have however to acknowledge the
development of linear scaling methods that allow obtaining the electronic energy
of quite large molecular systems [14–17]. Nevertheless, one cannot, for the time
being, carry out millions of such calculations which are nonetheless required to
sample all the possible conformations, necessaries to properly describe highly flexible and dynamic systems like biomolecules. This sampling is generally realized
by means of Molecular Dynamics or Monte Carlo techniques using classical force
fields [18]. The main drawback of such simulations is that they cannot describe
electronic phenomena (chemical reactions, electronic transitions) since electrons
are only implicitly taken into account via empirical parameters. We then face an
ambiguous situation where Quantum Chemistry is required but cannot be applied.
The first solution was proposed by Warshel in 1976 [19] at the semi-empirical
level and by Rivail and Assfeld [20] 20 years later at the ab initio level of theory.
The seminal idea is to divide the large molecular system into two communicating
parts, one considered as the active part (where the electronic phenomenon takes
place) is small and is described with Quantum Mechanics (QM) methods, the other,
considered as the surroundings, contains the remaining thousands of atoms and is
then treated with Molecular Mechanics (MM) Force Fields. This is the principle of
the so-called QM/MM methods, which rely on the locality of the electronic phenomenon under investigation [9].
In regard of the large panel of available QM methods (HF, PM3, PBE0, MPn,
CCSD, CI, …) [21–23] and of current MM force fields (AMBER, CHARMM, UFF,
DREIDING, …) [24–30], it exist an impressive bestiary of QM/MM couplings
[31–47]. Although these trivial differences can have a non-negligible effect on the
potential applications one can theoretically decipher, they won’t be discussed in this
chapter. A contrario we will focus our discussion on the fundamental differences to
treat the physical and or chemical interactions between the two sub-systems, QM
and MM. Once the general review will be set in the next section, the Local SelfConsistent Field (LSCF) method developed in our group [9, 20] will be detailed in
Sect. 1.3. Finally several illustrative examples will be given in the fourth section in
order to show the applicability and potency of the method.
A. Monari and X. Assfeld
is also of seminal importance in the growing field of phototherapy [5]. For instance
small organic or organometallic drugs can interact with DNA and induce permanent
lesions of the nucleic acid once activated by exposition to light [6–8], so as to
constitute very efficient photo-chemo-therapeutic agents. Taking care of the environment in theoretical calculations, both for ground and excited states is thus mandatory if one wants to obtain realistic results, i.e. directly comparable to experimental data, or to make reliable predictions [9–13].
The surrounding, i.e. the solvent or the bio-macromolecule, is a very large system, containing at least thousands of atoms to obtain an acceptable representation.
Even if the progresses of quantum chemistry over the past decades are tremendous (algorithms, computer, new method, …), describing such large systems with
quantum methods is still far beyond our computational capacities, especially when
dealing with electronically excited states. We have however to acknowledge the
development of linear scaling methods that allow obtaining the electronic energy
of quite large molecular systems [14–17]. Nevertheless, one cannot, for the time
being, carry out millions of such calculations which are nonetheless required to
sample all the possible conformations, necessaries to properly describe highly flexible and dynamic systems like biomolecules. This sampling is generally realized
by means of Molecular Dynamics or Monte Carlo techniques using classical force
fields [18]. The main drawback of such simulations is that they cannot describe
electronic phenomena (chemical reactions, electronic transitions) since electrons
are only implicitly taken into account via empirical parameters. We then face an
ambiguous situation where Quantum Chemistry is required but cannot be applied.
The first solution was proposed by Warshel in 1976 [19] at the semi-empirical
level and by Rivail and Assfeld [20] 20 years later at the ab initio level of theory.
The seminal idea is to divide the large molecular system into two communicating
parts, one considered as the active part (where the electronic phenomenon takes
place) is small and is described with Quantum Mechanics (QM) methods, the other,
considered as the surroundings, contains the remaining thousands of atoms and is
then treated with Molecular Mechanics (MM) Force Fields. This is the principle of
the so-called QM/MM methods, which rely on the locality of the electronic phenomenon under investigation [9].
In regard of the large panel of available QM methods (HF, PM3, PBE0, MPn,
CCSD, CI, …) [21–23] and of current MM force fields (AMBER, CHARMM, UFF,
DREIDING, …) [24–30], it exist an impressive bestiary of QM/MM couplings
[31–47]. Although these trivial differences can have a non-negligible effect on the
potential applications one can theoretically decipher, they won’t be discussed in this
chapter. A contrario we will focus our discussion on the fundamental differences to
treat the physical and or chemical interactions between the two sub-systems, QM
and MM. Once the general review will be set in the next section, the Local SelfConsistent Field (LSCF) method developed in our group [9, 20] will be detailed in
Sect. 1.3. Finally several illustrative examples will be given in the fourth section in
order to show the applicability and potency of the method.