comparison size_distributions.py @ 4:f1eeaf42144b draft default tip

planemo upload for repository https://github.com/bardin-lab/smallRNA_tools commit c8e0a703fcdff580ba0a0c5806a37c088c03ab7b-dirty

3:21b5a9170b90

                    reference_counters[reference]['antisense'][readlength] += 1
                else:
                    reference_counters[reference]['sense'][readlength] += 1
    df = pd.Panel(reference_counters).to_frame()
    df.index.names = ['readlength', 'orientation']
    return df


def to_long(df):
    df = df.reset_index()
    df = df.melt(id_vars=('readlength', 'orientation'))
    df.columns = ['readlength', 'orientation', 'reference', 'count']
    return df


def write_table(df, output_path):
    df.to_csv(output_path, sep="\t", index=False)

@click.option('--wide/--long', default=False, help="Output wide or long format.")
@click.option('--output', default="/dev/stdout", help="Write to this file")
def size_dist(alignment_path, minimum_size=18, maximum_size=30, output="/dev/stdout", wide=False):
    """Calculate size distribution and orientation"""
    with pysam.AlignmentFile(alignment_path) as alignment_file:
        df = global_size_distribution(alignment_file, minimum_size, maximum_size)
        if not wide:
            df = to_long(df)
        write_table(df, output)


if __name__ == '__main__':
    size_dist()

4:f1eeaf42144b

                    reference_counters[reference]['antisense'][readlength] += 1
                else:
                    reference_counters[reference]['sense'][readlength] += 1
    df = pd.Panel(reference_counters).to_frame()
    df.index.names = ['readlength', 'orientation']
    return i, df


def to_long(df, total_count):
    df = df.reset_index()
    df = df.melt(id_vars=('readlength', 'orientation'))
    df.columns = ['readlength', 'orientation', 'reference', 'count']
    df['TPM'] = df['count'] / total_count * 10 ** 6
    return df


def write_table(df, output_path):
    df.to_csv(output_path, sep="\t", index=False)

@click.option('--wide/--long', default=False, help="Output wide or long format.")
@click.option('--output', default="/dev/stdout", help="Write to this file")
def size_dist(alignment_path, minimum_size=18, maximum_size=30, output="/dev/stdout", wide=False):
    """Calculate size distribution and orientation"""
    with pysam.AlignmentFile(alignment_path) as alignment_file:
        total_count, df = global_size_distribution(alignment_file, minimum_size, maximum_size)
        if not wide:
            df = to_long(df, total_count)
        write_table(df, output)


if __name__ == '__main__':
    size_dist()