Define file inputs. Alternatively subclass typing.io.TextIO to be okay with sdin/stdout

from path import Path
class Fastq(Path): pass
class Fasta(Path): pass 
PairedEnd = Tuple[Fastq, Fastq]

Define options

from typing import *
#sort of like an Enums in other langauges
MiSeq = NamedTuple("MiSeq", [])
Roche454 = NamedTuple("Roche454", [])
IonTorrent = NamedTuple("IonTorrent", []) 

Platform = Union[MiSeq,Roche454,IonTorrent]

TrimOpts = NamedTuple('TrimOpts', 
        [('paired',bool), 
         ('trim_n', bool),
         ('q', Optional[int]),
         ('removebases', Optional[int]),
         ('platforms', List[Platform])])

Define some functions to process the files based on the options.

def trim_reads(f1: Fastq, f2: Fastq, opts: TrimOpts) -> PairedEnd:
    pass # do stuff with input

def fasta_to_fastq(fs: Fasta) -> Fastq:
    pass # do stuff with input

Run

if __name__ == '__main__':
    command = basic_command(trim_reads)
    command.run()
    command = command.sub_commands(trim_reads, prep_fastq)
    command.run()

Usage statement

<Fastq> <Fastq> --platforms ( <MiSeq> | <Roche454> | <IonTorrent> )... --trim_n  [ --removebases <int> ] --paired  --adapters <str>... [ --q <int> ]

Commandline Parsing

Works if you wrap everything that's required in the NamedTuple. Combining with positional arguments not yet working.

###TODO: It should be possible to automate commandline testing using hypotypes + typarser. All possible commandline-combinations would get run and you could check the properties one at a time, i.e., --csv flag means output should be in csv format, etc. This could report the exact commandline string.

parse/unparse tests for the parser/usage generator

typesafe pipelines could be created by matching up the types of inputs (using ADTs, e.g. CSVFile, etc.) to form the DAG.