0
$\begingroup$

I have a heatmap image (correlation between all matrix columns) and I'm straggling to preform all the changes below within the same image:

  1. bar colors should be symmetric around zero (e.g., correlation of 1 and -1 should be with the same color)
  2. change the correlation matrix to a diagonal matrix, since correlation values are symmetric - and show only upper matrix triangle (mask out the lower triangle )
  3. show the correlation values in every cell of the diagonal matrix
  4. x,y axis ticks - show the column names (instead of a serial number)

This is the code:

def generate_heatmap(X):

    """
    Pearson Correlation Heatmap Plot

    :return:
    """
    print("Start Pearson Correlation Heatmap Plot  .. ", datetime.now())

    plt.figure(figsize=(10,8))
    plt.title('Pearson Correlation of miRNAs', y=1.05, size=15)

    # Correlation matrix for heatmap
    corr = np.corrcoef(X.transpose())

    plt.imshow(corr, cmap='BuPu', interpolation='nearest')
    plt.colorbar()
    plt.show()
$\endgroup$
0
$\begingroup$

This is how I obtained the desired plot:

def generate_heatmap(X):

    """
    Pearson Correlation Heatmap Plot

    :return:
    """
    #from matplotlib import cm as CM
    from matplotlib.colors import LinearSegmentedColormap

    print("Start Pearson Correlation Heatmap Plot  .. ", datetime.now())

    # get column names
    cols = X.columns
    # define plot for heatmap
    fig, ax = plt.subplots(figsize=(16,16))

    # ------------------------------------------------------------
    # Correlation matrix for heatmap. the tranpose is because we want pxp matrix (rather a nxn)
    corr = np.corrcoef(X, rowvar=False)
    # show only upper matrix triangle - mask out the lower triangle of corr data
    corr = np.triu(corr, k=0) 

    # ------------------------------------------------------------
    # Edit graphics of the plot
    plt.title('Pearson Correlation of ' + str(len(cols)) + ' miRNAs', y=1.05, size=15, fontsize=32)

    # bar colors shold be symetric around zero!
    colors = [(1, 0, 0), 'w', (1, 0, 0)] 
    cm = LinearSegmentedColormap.from_list('heatmap', colors, N=20)    

    # ------------------------------------------------------------
    # Heatmap based on corr matrix we provided
    c = plt.pcolor(corr, edgecolors='w', linewidths=2, cmap=cm, vmin=-1.0, vmax=1.0)

    # ------------------------------------------------------------
    # Editing additional graphics of the plot (if not too big)
    if len(cols) < 50:

        # set axis label names
        ax.set_xticks(np.arange(len(cols)))
        ax.set_xticklabels(labels = cols, rotation=45, fontsize=12, ha='center')

        ax.set_yticks(np.arange(len(cols)))
        ax.set_yticklabels(labels = cols, rotation=45, fontsize=12)

        # show corr values in every cell
        for (i, j), z in np.ndenumerate(corr):
            # in the symetric values, don't annotate the cell with the corr value
            if (i > j): 
                continue
            else:
                # va and ha not working, we do +0.4 to overcome the centering of values 
                ax.text(j+0.4, i+0.4, '{:0.2f}'.format(z), ha='center', va='bottom', fontsize=11)

    plt.colorbar(c)
    plt.show()

#generate_heatmap(miRNA_data[selected_mir_columns])
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.