Unable to get coloured graph.

[varundewaker]

Hello chemosim-lab
When I run this code i get graph but it is not colored whereas legends are indicated in colored. I am using Ubuntu 20.04.6 LTS and notebook in firefox. I usually get following warning.
home/XXXX/miniconda3/envs/prolif/lib/python3.11/site-packages/seaborn/axisgrid.py:118: UserWarning: Tight layout not applied. tight_layout cannot make axes width small enough to accommodate all axes decorations
self._figure.tight_layout(*args, **kwargs). Please suggest how to get colored graph. Thanks in advance.

import seaborn as sns
import pandas as pd

reorganize data

data = df.reset_index()
data = pd.melt(data, id_vars=["Frame"], var_name=["residue","interaction"])
data = data[data["value"] != False]
data.reset_index(inplace=True, drop=True)

plot

sns.set_theme(font_scale=.8, style="white", context="talk")
g = sns.catplot(
data=data, x="interaction", y="Frame", hue="interaction", col="residue",
hue_order=["Hydrophobic", "HBDonor", "HBAcceptor", "PiStacking", "CationPi", "Cationic"],
height=3, aspect=0.2, jitter=0, sharex=False, marker="_", s=8, linewidth=3.5,
)
g.set_titles("{col_name}")
g.set(xticks=[], ylim=(-.5, data.Frame.max()+1))
g.set_xticklabels([])
g.set_xlabels("")
g.fig.subplots_adjust(wspace=0)
g.add_legend()
g.despine(bottom=True)
for ax in g.axes.flat:
ax.invert_yaxis()
ax.set_title(ax.get_title(), pad=15, rotation=60, ha="center", va="baseline")

[cbouy]

Hi,

What do you mean by "not colored", is it just one single color or you can't see anything at all?
Also, what's the output of df.melt()["interaction"].value_counts()?

Best,
Cédric

[varundewaker]

Thanks for your response. This is the part of output file. Hope you understand the issue clearly now. I used this command
df = fp.to_dataframe()
df
df.melt()["interaction"].value_counts()

The output I get
VdWContact 4422
Hydrophobic 4020
HBAcceptor 603
PiCation 603
HBDonor 603
PiStacking 402
Name: interaction, dtype: int64

[cbouy]

Okay I see now! Which version of seaborn are you using? You can get the version with:

import seaborn as sns
print(sns.__version__)

It might be that there's an edge around the markers in the plot which hides the actual color due to recent changes in the library

[varundewaker]

I got this when I executed this command in jupyter
import seaborn as sns
print(sns.version)

0.12.2

[cbouy]

Try installing 0.11.0 instead:
pip install seaborn==0.11.0 or conda install seaborn==0.11.0 depending on which package manager you use

[varundewaker]

This frozen installation of conda install seaborn==0.11.0 didn't worked for me because it downgraded other libraries also. I used old one https://github.com/chemosim-lab/ProLIF-paper old version and its working.
During installation i think https://github.com/chemosim-lab/ProLIF/blob/master/.binder/environment.yml, the given seaborn is not installing rather it install updated seaborn package (0.12.2). Please look in this matter. Thanks for your support.

[cbouy]

I see, I plan to include this seaborn code snippet directly in Prolif for the next update so I'll address this compatibility issue by then.