Heatmap visualisations for quantified (virus) metagenomics classifications

In [1]:

    

#Import all required libraries
import numpy as np
import pandas as pd
from bokeh.plotting import figure, show, output_file
from bokeh.models import HoverTool, ColumnDataSource
from bokeh.io import output_notebook
import glob


    

In [2]:

    

output_notebook()


    

    






    

        
        Loading BokehJS ...
    






    

In [3]:

    

#Set the variables.
FOLDER = "../results/"
FILE_WILDCARD = "*_results.csv"

COLUMN_CLASSIFICATION = "Assignment"
COLUMN_LOAD = "Mapped # Reads"

METHOD = "Genome Detective"

#Make sure to adjust these values to a size that fits your data.
HEATMAP_TITLE = "Viruses classified by %s" % METHOD
HEATMAP_WIDTH = 700
HEATMAP_HEIGHT = 500
HEATMAP_TITLE_FONT_SIZE = "16pt"
HEATMAP_AXIS_FONT_SIZE = "12pt"

COLOUR = ["#6b2d18"] #Selected from coffee beans: http://s.eatthis-cdn.com/media/images/ext/851818315/coffee-beans.jpg

#This is where you can save the resulting html file.
#Write the directory path + file name you want to use,
# and the title you want the file to have.
#This title will appear in the top bar of your browser window.
OUTPUT_FILE = "../results/test_heatmap.html"
OUTPUT_TITLE = "Heatmap of viruses classified by %s" % METHOD


    

In [4]:

    

#Define the required functions
def create_concatenated_dataframe(file_wildcard, folder):
    """
    Input: a filename with wildcard,
            and a folder name
    Output: One concatenated dataframe of all the input files
    """
    #Step 1: create a list of the files using glob
    file_list = glob.glob("%s%s" % (folder, file_wildcard))
    file_list = sorted(file_list)
    
    file_suffix = file_wildcard.lstrip("*")
    
    #Step 2: open the files as dataframe, remove "Contigs" column and add sample IDs
    
    df_list = []
    for results_file in file_list:
        results_df = pd.read_csv(results_file)
        results_df = results_df.drop("Contigs", axis = 1) #remove unnecessary (and long!) column
        sample_id = results_file[:results_file.index(file_suffix)].split('/')[-1]
        results_df["sample"] = sample_id
        df_list.append(results_df)

    #Step 3: concatenate the dataframes
    super_df = pd.concat(df_list, ignore_index=True)
    
    return(super_df)

def create_and_show_heatmap(dataframe, column_classification, column_load):
    """
    Input: Dataframe, names of columns:
        'classification'
        'load'
    Output: heatmap (shown in the notebook and 
            exported to the html file defined as variable above)
    """
    samples = dataframe["sample"]
    assignments = dataframe[column_classification]
    loads = dataframe[column_load]
    #a fix for numbers with tildes, e.g. '~80'
    loads_new = []
    for value in loads:
        if isinstance(value, str):
            value = value.strip('~')
        loads_new.append(int(value))
    
    colors = len(loads) * COLOUR #multiply to make an equally long list
    
    max_load = max(loads_new)
    alphas = [ min( x / float(max_load), 0.9) + 0.1 for x in loads_new ]
    
    source = ColumnDataSource(
            data = dict(samples=samples, assignments=assignments, colors=colors, loads=loads, alphas=alphas)
        )

    TOOLS = "hover, save, pan, box_zoom, wheel_zoom, reset"

    p = figure(title = HEATMAP_TITLE,
              x_range = list(sorted(set(samples))),
              y_range = list(reversed(sorted(set(assignments)))), #reverse to order 'from top to bottom'
              x_axis_location = "above",
              toolbar_location="right",
              tools = TOOLS)

    p.plot_width = HEATMAP_WIDTH
    p.plot_height = HEATMAP_HEIGHT
    p.grid.grid_line_color = None
    p.axis.axis_line_color = None
    p.axis.major_tick_line_color = None
    p.axis.major_label_text_font_size = HEATMAP_AXIS_FONT_SIZE
    p.axis.major_label_standoff = 0
    p.xaxis.major_label_orientation = np.pi/4
    p.title.text_color = COLOUR[0]
    p.title.text_font_size = HEATMAP_TITLE_FONT_SIZE

    p.rect("samples", "assignments", 1, 1, source=source,
           color="colors", alpha="alphas", line_color=None)

    p.select_one(HoverTool).tooltips = [
        ('Sample', "@samples"),
        ('Taxon' , "@assignments"),
        ('Number of reads', "@loads"),
    ]
    
    output_file(OUTPUT_FILE, title=OUTPUT_TITLE)
    
    show(p)


    

In [5]:

    

df = create_concatenated_dataframe(file_wildcard = FILE_WILDCARD, folder = FOLDER)

create_and_show_heatmap(dataframe=df,
                        column_classification=COLUMN_CLASSIFICATION,
                        column_load=COLUMN_LOAD)