Example files used while developing pyqi's Getting Started tutorials.

README.md

These files were used while developing pyqi's Getting Started tutorials. See those documents for usage examples.

seqs.fna

>s1
ACCTTTAACC
>s2
CCGG
>s3
AAAAAAAAAAAAAAAAAAAAAAAAAAA

sequence_collection_summarizer.py

#!/usr/bin/env python

from __future__ import division
from pyqi.core.command import Command, Parameter, ParameterCollection

__author__ = "Greg Caporaso"
__copyright__ = "Copyright 2013, Greg Caporaso"
__credits__ = ["Greg Caporaso"]
__license__ = "BSD"
__version__ = "0.0.1"
__maintainer__ = "Greg Caporaso"
__email__ = "[email protected]"

class SequenceCollectionSummarizer(Command):
    BriefDescription = "Generate summary statistics on a collection of sequences."
    LongDescription = "Provided the number of sequences, the minimum sequence length, and the maximum sequence length given a collection of sequences. Sequences should be provided as a list (or generator) of tuples of (sequence id, sequence) pairs."
    Parameters = ParameterCollection([
        Parameter(Name='seqs', DataType=list,
                  Description='sequences to be summarized', Required=True),
        Parameter(Name='suppress_length_summary', DataType=bool,
                  Description='do not generate summary information on the sequence lengths', 
                  Required=False,Default=False)
    ])

    def run(self, **kwargs):
        """
        """
        num_seqs = 0
        sequence_lengths = []
        for seq_id, seq in kwargs['seqs']:
            num_seqs += 1
            sequence_lengths.append(len(seq))
            
        if kwargs['suppress_length_summary']:
            min_length = None
            max_length = None
        else:
            min_length = min(sequence_lengths)
            max_length = max(sequence_lengths)
        
        return {'num-seqs':num_seqs,
                'min-length':min_length,
                'max-length':max_length}

CommandConstructor = SequenceCollectionSummarizer

summarize_sequence_collection.py

#!/usr/bin/env python

from pyqi.core.interfaces.optparse import (OptparseUsageExample,
                                           OptparseOption, OptparseResult)
from pyqi.core.command import make_parameter_collection_lookup_f
from sequence_collection_summarizer import CommandConstructor
from pyqi.core.exception import IncompetentDeveloperError
import os

param_lookup = make_parameter_collection_lookup_f(CommandConstructor)

def parse_fasta(fp):
    """
       fp: path to a fasta-formatted file
       
       This function is a fasta record generator, yielding 
        (sequence id, sequence) pairs when provided with a 
        valid fasta file.
       
       NO ERROR CHECKING IS PERFORMED!
    """
    # Always open files for reading in python using mode 'U'
    # to correctly handle different types of line breaks
    f = open(fp,'U')
    seq_id = None
    seq = []
    for line in f:
        line = line.strip()
        if line.startswith('>'):
            if len(seq) != 0:
                # we've completed a fasta record
                yield seq_id, ''.join(seq)
            seq_id = line[1:]
            seq = []
        else:
            seq.append(line)
    yield seq_id, ''.join(seq)

def append_datum_to_file(result_key, data, option_value=None):
    """Append summary information to a file.
    """
    # don't do anything if data is None
    if data is None:
        return
    
    # If option_value is None when this output handler is called, 
    # the interface developer did something wrong when defining
    # the OptparseResults. Politely alert the developer that
    # this output handler isn't associated with an option
    # (it needs to be associated with an output file path).
    if option_value is None:
        raise IncompetentDeveloperError(
         "Cannot write output without a filepath.")
    
    # open the output file for appending, and write the 
    # summary information to a single tab-separated line
    with open(option_value, 'a') as f:
        f.write('%s\t%d\n' % (result_key, data))

usage_examples = [
    OptparseUsageExample(ShortDesc="Summarize the input sequence collection and write the result to file.",
                         LongDesc="Read the file specified by -i, and compute the number of sequences in the file, and the minimum and maximum sequence lengths. Write all of that information to path specified by -o.",
                         Ex="%prog -i seqs.fna -o seqs.summary.txt"),
    OptparseUsageExample(ShortDesc="Summarize the input sequence collection and write the result to file, excluding information on sequence lengths.",
                         LongDesc="Read the file specified by -i, compute the number of sequences in the file, and write that information to path specified by -o.",
                         Ex="%prog -i seqs.fna -o seqs.summary.txt --suppress-length-summary")
]

inputs = [

    OptparseOption(Parameter=param_lookup('seqs'),
                   InputType='existing_filepath',
                   InputAction='store',
                   InputHandler=parse_fasta,
                   ShortName='i'),
                   
    OptparseOption(Parameter=param_lookup('suppress_length_summary'),
                   InputType=None,
                   InputAction='store_true',
                   InputHandler=None,
                   ShortName=None),

    OptparseOption(Parameter=None,
                   InputType='new_filepath',
                   InputAction='store',
                   ShortName='o',
                   Name='output-fp',
                   Required=True,
                   Help='path where output should be written')
]

outputs = [
    OptparseResult(ResultKey='num-seqs',
                   OutputHandler=append_datum_to_file,
                   OptionName='output-fp'), 
    OptparseResult(ResultKey='min-length',
                   OutputHandler=append_datum_to_file,
                   OptionName='output-fp'), 
    OptparseResult(ResultKey='max-length',
                   OutputHandler=append_datum_to_file,
                   OptionName='output-fp'), 

]