Publication

array(14 items)
   uid => 964 (integer)
   title => 'Crossover Operators for Molecular Graphs with an Application to Virtual Drug
       Screening' (86 chars)
   abstract => 'Genetic Algorithms are a powerful method to solve optimization problems with
       complex cost functions over vast search spaces that rely in particular on r
      ecombining parts of previous solutions. Crossover operators play a crucial r
      ole in this context. Here, we describe a large class of these operators desi
      gned for searching over spaces of graphs. These operators are based on intro
      ducing small cuts into graphs and rejoining the resulting induced subgraphs 
      of two parents. This form of cut-and-join crossover can be restricted in a c
      onsistent way to preserve local properties such as vertex-degrees (valency),
       or bond-orders, as well as global properties such as graph-theoretic planar
      ity. In contrast to crossover on strings, cut-and-join crossover on graphs i
      s powerful enough to ergodically explore chemical space even in the absence 
      of mutation operators. Extensive benchmarking shows that the offspring of mo
      lecular graphs are again plausible molecules with high probability, while at
       the same time crossover drastically increases the diversity compared to ini
      tial molecule libraries. Moreover, desirable properties such as favorable in
      dices of synthesizability are preserved with sufficient frequency that candi
      date offsprings can be filtered efficiently for such properties. As an appli
      cation we utilized the cut-and-join crossover in REvoLd, a GA-based system f
      or computer-aided drug design. In optimization runs searching for ligands bi
      nding to four different target proteins we consistently found candidate mole
      cules with binding constants exceeding the best known binders as well as can
      didates found in make-on-demand libraries. Taken together, cut-and-join cros
      sover operators constitute a mathematically simple and well-characterized ap
      proach to recombination of molecules that performed very well in real-life C
      ADD tasks.' (1834 chars)
   authors => array(8 items)
      0 => array(3 items)
         last_name => 'Domschke' (8 chars)
         first_name => 'Nico' (4 chars)
         sorting => 1 (integer)
      1 => array(3 items)
         last_name => 'Schmidt' (7 chars)
         first_name => 'Bruno J.' (8 chars)
         sorting => 2 (integer)
      2 => array(3 items)
         last_name => 'Gatter' (6 chars)
         first_name => 'Thomas' (6 chars)
         sorting => 3 (integer)
      3 => array(3 items)
         last_name => 'Golnik' (6 chars)
         first_name => 'Richard' (7 chars)
         sorting => 4 (integer)
      4 => array(3 items)
         last_name => 'Eisenhuth' (9 chars)
         first_name => 'Paul' (4 chars)
         sorting => 5 (integer)
      5 => array(3 items)
         last_name => 'Liessmann' (9 chars)
         first_name => 'Fabian' (6 chars)
         sorting => 6 (integer)
      6 => array(3 items)
         last_name => 'Meiler' (6 chars)
         first_name => 'Jens' (4 chars)
         sorting => 7 (integer)
      7 => array(3 items)
         last_name => 'Stadler' (7 chars)
         first_name => 'Peter Florian' (13 chars)
         sorting => 8 (integer)
   type => '1' (1 chars)
   keywords => '' (0 chars)
   year => 2024 (integer)
   affiliation => 0 (integer)
   link_paper => '' (0 chars)
   link_supplements => '' (0 chars)
   file_published => 0 (integer)
   journal => 'ChemRxiv' (8 chars)
   doi => '10.26434/chemrxiv-2024-41295' (28 chars)
   preprint => '-1' (2 chars)