Peptide Design Genetic Algorithm (PDGA)

run python run_PDGA.py population-size mutation-rate generation-gap query similarity-threshold topology

e.g. run_PDGA.py 50 1 0.8 Ala-Leu-Cys1-His-Gaba-Cys1-Ile 300 cyclic where:

• each generation will have 50 individuals
• 80% (0.80) of the individuals will be replaced, and 100% (1) of the new ones will be mutated
• PDGA will try to reach the MXFP value of the query Ala-Leu-Cys1-His-Gaba-Cys1-Ile
• compounds with CBD smaller than 300 from Ala-Leu-Cys1-His-Gaba-Cys1-Ile will be annotated in a results file
• the generated sequences will have cyclic topology.

topology can be linear, cyclic, or dendritic.

Sequences building blocks:

• the 20 natural amino acids as three-letters code (e.g. Ala)
• Orn Ornithine
• Hyp Hydroxyproline
• bAla Beta-alanine
• Gaba Gamma-aminobutyric acid
• a5a Delta-aminopentanoic acid
• a6a Epsilon-aminohexanoic acid
• a7a Zeta-aminoheptanoic acid
• a8a Eta-aminooctanoic acid
• a9a Theta-aminononaanoic acid
• Dap 2,3-diaminopropionic acid as branching unit
• Dab 2,4-diaminobutyric acid as branching unit
• BOrn Ornithine as branching unit
• BLys Lysine as branching unit
• cy Head-to-tail cyclization. It is always placed at the beginning (left, N terminus) of the sequence.
• Cys1 First pair of cyclizes cysteines. Always in pair, never next to each other.
• Cys2 Second pair of cyclizes cysteines. They are always present in pair, never next to each other, present only if Cys1 is already part of the sequence.
• Cys3 Third pair of cyclizes cysteines. They are always present in pair, never next to each other, present only if Cys1 and Cys2 are already part of the sequence.
• Ac N-terminus acetylation. It is always placed at the beginning (N-terminus, left) of the sequence
• NH2 C-terminus amide. It is always placed at the end (C-terminus, right) of the sequence

By default, all building blocks are used. Using the method exclude_buildingblock(bb) is possible to exclude all building blocks with the exception of the head to tail cyclization (cy) and the cyclized cysteines (Cys1, Cys2, Cys3), to exclude cyclization use “topology = linear”.

By default, PDGA stops after CBD = 0 from the query MXFP value is found 10 times. Using the method ga.set_time_limit('hh:mm:ss') is possible to stop PDGA after a chosen time.

run_PDGA.py uses class PDGA in PDGA_class.py.

To run the genetic algorithm with MXFP as fitness function, a valid Chemaxon licence is required and the specified libraries needs to be downloaded.