Plot most common barcodes

plot_the_percentages.py

from __future__ import with_statement
import matplotlib.pyplot as plt
plt.rc('xtick', labelsize='x-small')
import yaml


def convert(yaml_file):
    """Convert the data in a yaml file generated by count_barcodes.py to a
    .dat file with space seperated sorted fractions of barcode distribution.
    """
    yaml_yaml = open(yaml_file)
    yaml_data = yaml.load(yaml_yaml)
    dat_data = open(yaml_file.split(".")[0] + ".dat", "w+")

    matched_bc_and_counts = {}
    for barcode, info in yaml_data["matched"].items():
        matched_bc_and_counts[barcode] = info["count"]
    unmatched_bc_and_counts = {}
    for barcode, info in yaml_data["unmatched"].items():
        unmatched_bc_and_counts[barcode] = info["count"]

    total = float(sum(matched_bc_and_counts.values()) + \
    sum(unmatched_bc_and_counts.values()))
    matched_sorted_counts = [(barcode, count / total)  \
    for barcode, count in matched_bc_and_counts.iteritems()]
    matched_sorted_counts.sort(key=lambda x: x[1], reverse=True)
    unmatched_sorted_counts = [(barcode, count / total) \
    for barcode, count in unmatched_bc_and_counts.iteritems()]
    unmatched_sorted_counts.sort(key=lambda x: x[1], reverse=True)
    print total

    yaml.dump({'matched': matched_sorted_counts, \
    'unmatched': unmatched_sorted_counts}, dat_data)


def plot_dat(dat_file):
    """Plot the sorted lists of tuples.
    """
    fig = plt.figure()
    fig.subplots_adjust(bottom=0.2)
    ax = fig.add_subplot(111)

    data = open(dat_file, "r")
    tuple_lists = yaml.load(data)
    y_matched = [float(value) for _, value in tuple_lists["matched"]]
    y_unmatched = [float(value) for _, value in tuple_lists["unmatched"]]
    top_num = min(40, len(y_matched) + len(y_unmatched))
    percentile = 0.02

    ax.bar(range(len(y_matched)), y_matched, color='green')
    ax.bar(range(len(y_matched), top_num), \
    y_unmatched[:top_num - len(y_matched)], color='blue')
    ax.plot([0, top_num], [percentile, percentile], 'r-', linewidth=3)

    ticks = ['\n'.join(list(barcode)) + '\n#' for barcode, _ in \
    tuple_lists["matched"]]
    ticks += ['\n'.join(list(barcode)) for barcode, _ in \
    tuple_lists["unmatched"][:top_num - len(tuple_lists["matched"])]]
    plt.xticks([x + .4 for x in range(top_num)], ticks)

    fig.show()


def sort_out_N(yaml_file, out_yaml):
    """Goes through the matched barcodes, looking which ones contains 'N' and
    tries to replace that with a matched barcode which does not contain 'N'.
    If that is not possible, the barcode will be moved to the unmatched
    catergory.
    """
    with open(yaml_file) as yaml_yaml:
        yaml_data = yaml.load(yaml_yaml)

    for barcode, info in yaml_data["matched"].items():
        if 'N' in barcode:
            for matched_barcode in info["variants"]:
                if 'N' not in matched_barcode:
                    yaml_data["matched"][matched_barcode] = info
                    break
            else:
                yaml_data["unmatched"][barcode] = info

            del yaml_data["matched"][barcode]

    with open(out_yaml, "w+") as out_handle:
        yaml.dump(yaml_data, out_handle)