Color conditional data on a plot with matplotlib threw a loop

Question

I have a the following dataframe

import pandas as pd
import matplotlib.pyplot as plt

datas = [['RAC1','CD0287',1.52,9.88], ['RAC1','CD0695',2.08,10.05],['RAC1','CD0845',2.01,10.2], ['RAC3','CD0258',1.91,9.8], ['RAC3','CD471',1.66,9.6], ['RAC8','CD0558',1.32,9.3], ['RAC8','CD0968',2.89,10.01]]
labels = ['Plate', 'Sample', 'LogRatio', 'Strength']
df = pd.DataFrame(data = datas, columns=labels, index=[8, 3, 5, 4, 12, 44, 2])
print(df)

    Plate  Sample  LogRatio  Strength
8   RAC1  CD0287      1.52      9.88
3   RAC1  CD0695      2.08     10.05
5   RAC1  CD0845      2.01     10.20
4   RAC3  CD0258      1.91      9.80
12  RAC3   CD471      1.66      9.60
44  RAC8  CD0558      1.32      9.30
2   RAC8  CD0968      2.89     10.01

As you can see, my data are distributed on different plates. I would like to create as many plot as I have different plates : 3 plots. And on each plot, I would like to color one plate in red and the others in black.

The only way I found so far, is to do it manually by writing the code for each plate, and change the red plate for earch run (I have more than 30 plates in reality so it takes too many time). I can still show you my code if it can help you to understand:

def getIndexPlates(df):
    listIndicesAllPlates = []
    df = df.reset_index()
    for name,group in df.groupby("Plate"): 
        temp_list = []
        temp_list.append(name)
        temp_list.append(group.index.tolist()) #create a tuple with the name of the plate and the index of all the samples in this plate
        listIndexAllPlates.append(temp_list)
    return listIndexAllPlates

def plotting(df,listIndexAllPlates): 
    plt.clf()
    ax=plt.gca()
    datas = df[["LogRatio", "Strength"]].as_matrix()
    for sample in range(len(datas)):
        if sample in listIndexAllPlates[0][1]: #if the sample is on the the first tuple of my list -> on the first plate
            ax.scatter(datas[sample,0], datas[sample,1], alpha=0.8, facecolors='none', edgecolors='red')
        if sample in listIndexAllPlates[1][1]:
            ax.scatter(datas[sample,0], datas[sample,1], alpha=0.8, facecolors='none', edgecolors='black')
        if sample in listIndexAllPlates[2][1]:
            ax.scatter(datas[sample,0], datas[sample,1], alpha=0.8, facecolors='none', edgecolors='black')
    plt.show()

listIndexAllPlates = getIndexPlates(df)
plotting(df,listIndexAllPlates)

So here I have my first plot with the plate 'RAC1' in red and RAC3 and RAC8 in black, and now I would like to have the second plot with RAC3 in red (RAC1 and RAC8 in black) and the third plot with RAC8 in red (RAC1 and RAC3 in black). To do so I manually change the color in my function but I would like a solution to do it automatically. And I know my way is really a bad and ugly way, I just don't know how to do it.

Answer 1

You can use groupby here in conjunction with difference of the pandas Index object to loop through your plates and get the indices for the current plate and the rest of them:

for label, plate_df in df.groupby("Plate"):
    plate_indices = plate_df.index
    rest_indices = df.index.difference(plate_indices)

    # do your plotting here accordingly

    print(label, plate_indices, rest_indices)

RAC1 Int64Index([8, 3, 5], dtype='int64') Int64Index([2, 4, 12, 44], dtype='int64')
RAC3 Int64Index([4, 12], dtype='int64') Int64Index([2, 3, 5, 8, 44], dtype='int64')
RAC8 Int64Index([44, 2], dtype='int64') Int64Index([3, 4, 5, 8, 12], dtype='int64')

Edit

To include the plotting, just include your matplotlib statements:

plot_kwargs = {"alpha": 0.8, "facecolors": "none"}
for label, plate_df in df.groupby("Plate"):
    plate_indices = plate_df.index
    rest_indices = df.index.difference(plate_indices)

    # create plot
    plt.clf()
    ax=plt.gca()
    ax.scatter(df.loc[plate_indices, "LogRatio"], df.loc[plate_indices, "Strength"], edgecolors='red', **plot_kwargs)
    ax.scatter(df.loc[rest_indices, "LogRatio"], df.loc[rest_indices, "Strength"], edgecolors='black', **plot_kwargs)
    plt.show()