Scatter plot for non-numeric DataFrame

user_13014274 0
Viewed 164

I have a DataFrame that consists only of non-numeric data. I have two exemplary sequences:

seq_1 = 'AATGMAM'
seq_2 = 'TATAMTM'

Where one of these sequences is used as columns, and the second one as the index. If the letter overlaps, the DataFrame is filled with a '*' sign. Otherwise, the value is ''.

data = [["*" if p1 == p2 else "" for p2 in seq_2] for p1 in seq_1]
df = pd.DataFrame(data, columns=list(seq_1), index=list(seq_2)

  A A T G M A M
T     *        
A * *       *
T     *
A * *       *
M
T     *
M         *   *

Now I want to create a scatter plot that would depict this DataFrame. The x-axis should be the index and y-axis the columns. How can I do that?

EDIT: Thanks to @Shaido, I was able to plot it. However I have issues with adding a separate color for each label.

color_keys = max([np.unique(df.index), np.unique(df.columns)], key=len)
rgb_values = sns.color_palette("Set2", len(color_keys))
colors = dict(zip(color_keys, rgb_values))

for g in np.unique(df.index):
    ix = np.where(df.columns == g)
    plt.scatter(x[ix], y[ix], c = colors[g], label = g, s = 100)

plt.xticks(np.arange(df.shape[0]), df.index)
plt.yticks(np.arange(df.shape[1]), df.columns)
plt.legend()

plt.show()

How can I map each unique value to a separate column?

scatter pandas matplotlib plot python
1 Answers

You can get the coordinates for the scatter plot using np.where, after that we adjust the y values to reverse the axis as follows:

y, x = np.where(df == "*")
y = len(seq_2) - y

Now we can plot it, not that the y-ticks are reversed as well:

fig, ax = plt.subplots()
ax.scatter(x, y)
ax.set_xticklabels(' ' + seq_1)
ax.set_yticklabels(' ' + seq_2[::-1])
plt.show()

Result:

enter image description here

user_75795470
Related