How do I log transform my y-axis in my plot? scale('log') does not work in pandas

Question

Here I'm making a scatterplot that contains data from 3 columns out of my dataframe. The columns are the different samples and row names are genes (thus the same for each col). I would like my y-axis to be log-transformed. However the standard way to use yscale('log') does not work. Some how it limits the y-axis, without making it logarithmic and data gets lost.

This is before:

ax = mini_df.plot(kind="scatter", x="gene",y="sample_02A", color="b", label="sample 02A")
    mini_df.plot(kind="scatter", x="gene",y="sample_04A", color="g", label="sample 04A", ax=ax)
    mini_df.plot(kind="scatter", x="gene",y="sample_06A", color="r", label="sample 06A", ax=ax)
    mini_df.plot(kind="scatter", x="gene",y="sample_08A", color="purple", label="sample 08A", ax=ax)
    ax.set_xlabel("gene")
    ax.set_ylabel("expression value")
    
    plt.show()

enter image description here

This is after logarithmic transformation

 ax = mini_df.plot(kind="scatter", x="gene",y="sample_02A", color="b", label="sample 02A")
mini_df.plot(kind="scatter", x="gene",y="sample_04A", color="g", label="sample 04A", ax=ax)
mini_df.plot(kind="scatter", x="gene",y="sample_06A", color="r", label="sample 06A", ax=ax)
mini_df.plot(kind="scatter", x="gene",y="sample_08A", color="purple", label="sample 08A", ax=ax)
ax.set_xlabel("gene")
ax.set_ylabel("expression value")
ax.set_yscale('log')

plt.show()

enter image description here

Did you research the documentation for DataFrame.plot()? pandas.pydata.org/docs/reference/api/pandas.DataFrame.plot.html — gshpychka
– gshpychka, Commented Aug 18, 2021 at 7:37
What should I do when someone answers my question? stackoverflow.com/help/someone-answers — Niv Dudovitch
– Niv Dudovitch, Commented Sep 2, 2021 at 13:25

Niv Dudovitch · Accepted Answer · 2021-08-18 10:02:47Z

Option 1:

you can just log the y-axis yourself before plotting it. For example: (you can modify all the columns to logarithmic together instead of do it 'column by column' like my example..(

mini_df_copy = mini_df.copy(deep=True)
mini_df_copy.sample_02A = np.log(mini_df_copy.sample_02A)
mini_df_copy.sample_04A= np.log(mini_df_copy.sample_04A)
mini_df_copy.sample_06A= np.log(mini_df_copy.sample_06A)
mini_df_copy.sample_08A= np.log(mini_df_copy.sample_08A)

ax = mini_df_copy.plot(kind="scatter", x="gene",y="sample_02A", color="b", label="sample 02A")
mini_df_copy.plot(kind="scatter", x="gene",y="sample_04A", color="g", label="sample 04A", ax=ax)
mini_df_copy.plot(kind="scatter", x="gene",y="sample_06A", color="r", label="sample 06A", ax=ax)
mini_df_copy.plot(kind="scatter", x="gene",y="sample_08A", color="purple", label="sample 08A", ax=ax)
ax.set_xlabel("gene")
ax.set_ylabel("expression value")

Option 2:

You can just specify the 'ylim' bounds and then no points will disappear. ax.set_ylim((1, "max point value you want to display"))

Option 3:

Use 'symlog' instead of 'log': ax.set_yscale('symlog')

According to the points with value 0: log(0) is undefined so yscale('log') modify it to None, Matplotlib just ignores the NaNs and doesn't visualize them.

Collectives™ on Stack Overflow

How do I log transform my y-axis in my plot? scale('log') does not work in pandas

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related