Skip to content

Commit

Permalink
Worked on Plot2DScatters (not yet complete though)
Browse files Browse the repository at this point in the history
  • Loading branch information
artuurC committed Sep 7, 2023
1 parent 5627553 commit a53f4a6
Show file tree
Hide file tree
Showing 2 changed files with 58 additions and 21 deletions.
67 changes: 46 additions & 21 deletions src/FlowSOM/plotting/plot_functions.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -3,12 +3,12 @@
import pandas as pd

import matplotlib.colors
import random

from ..tools import get_channels, get_markers
from matplotlib import collections as mc
from matplotlib import gridspec
from ._plot_helper_functions import *
from scipy.stats import gaussian_kde


def plot_2D_scatters(
Expand Down Expand Up @@ -66,33 +66,30 @@ def plot_2D_scatters(
assert (
"marker" in xy_labels or "channel" in xy_labels
), f'xy_labels should be a list containing "marker" and/or "channel".'
assert isinstance(clusters[0], list), f"clusters should be a list of lists."
assert isinstance(metaclusters[0], list), f"metaclusters should be a list of lists."
assert isinstance(channelpairs[0], list), f"channelpairs should be a list of lists."

metacluster = fsom.get_cell_data().obs["metaclustering"]

cell_metacluster = fsom.get_cell_data().obs["metaclustering"]
n_metaclusters = fsom.get_cell_data().uns["n_metaclusters"]
cell_cluster = fsom.get_cell_data().obs["clustering"]

bgI = random.sample(
range(fsom.get_cell_data().X.shape[0]), min([fsom.get_cell_data().X.shape[0], max_background_points])
metacluster = fsom.get_cluster_data().obs["metaclustering"]
bgI = np.random.choice(
range(fsom.get_cell_data().X.shape[0]),
min([fsom.get_cell_data().X.shape[0], max_background_points]),
replace=False,
)

if clusters is None:
clusters = []
if metaclusters is None:
metaclusters = []
# fig, axs = plt.subplots(len(channelpairs), len(clusters) + len(metaclusters))
fig = plt.figure()
spec = gridspec.GridSpec(ncols=len(channelpairs), nrows=len(clusters) + len(metaclusters))
subsets = {"Cluster": np.array(clusters), "Metacluster": np.array(metaclusters)}
for i, group in enumerate(subsets.keys()):
if len(subsets[group] == 0):
next
for j, subset in enumerate(subsets[group]):
"""
color += 1
"""
rowI = i + j
n = int(subset)

rowI = (i * len(subsets)) + j
n = [int(x) for x in subset]
for k, channelpair in enumerate(channelpairs):
channelpair = list(get_channels(fsom, channelpair).keys())
df_bg = np.array(fsom.get_cell_data().X[bgI, :])
Expand All @@ -103,11 +100,29 @@ def plot_2D_scatters(
df_ss = fsom.get_cell_data().X[clusters_OI, :]
df_ss = df_ss[:, indices_markers]
df_ss = np.c_[df_ss, cell_cluster[clusters_OI]]
col = np.asarray([gg_color_hue()(i) for i in range(n_metaclusters)])[
np.array([int(i) for i in metacluster])[n]
]
col_dict = {i: j for i, j in zip(np.unique(df_ss[:, 2]), col)}
cols = [col_dict[i] for i in df_ss[:, 2]]

df_c = np.c_[fsom.get_cluster_data()[n, indices_markers].X, n]
else:
df_ss = fsom.get_cell_data().X[np.isin(metacluster.astype(int), n), :]
metaclusters_OI = np.isin(cell_metacluster.astype(int), n)
df_ss = fsom.get_cell_data().X[metaclusters_OI, :]
df_ss = df_ss[:, indices_markers]

df_ss = np.c_[df_ss, cell_metacluster[metaclusters_OI]]
col = np.asarray([gg_color_hue()(i) for i in range(n_metaclusters)])[n]
col_dict = {i: j for i, j in zip(np.unique(df_ss[:, 2]), col)}
cols = [col_dict[i] for i in df_ss[:, 2]]
cl_in_mcl = np.where(np.isin(metacluster.astype(int), n))[0]
df_c = np.c_[fsom.get_cluster_data()[cl_in_mcl, indices_markers].X, cl_in_mcl]
plural = "s" if len(subset) > 1 else ""
cl_or_mcl = group + plural + ": "
subset_names = (
", ".join([str(i) for i in subset]) if group == "Cluster" else ", ".join([str(i) for i in subset])
)
title = cl_or_mcl + subset_names
if len(xy_labels) == 1 and xy_labels[0] == "channel":
xy_label = list(get_channels(fsom, channelpair).keys())
elif len(xy_labels) == 1 and xy_labels[0] == "marker":
Expand All @@ -116,12 +131,22 @@ def plot_2D_scatters(
channel_label = list(get_channels(fsom, channelpair).keys())
marker_label = list(get_markers(fsom, channelpair).keys())
xy_label = [marker_label[i] + " (" + channel_label[i] + ")" for i in range(len(channel_label))]
ssI = random.sample(range(df_ss.shape[0]), min([df_ss.shape[0], max_points]))
ssI = np.random.choice(range(df_ss.shape[0]), min([df_ss.shape[0], max_points]), replace=False)
df_ss = df_ss[ssI, :]
ax = fig.add_subplot(spec[rowI + (rowI + k)])
cols = np.array(cols)[ssI, :]
ax = fig.add_subplot(spec[rowI, k])
ax.scatter(df_bg[:, 0], df_bg[:, 1], c="grey", s=size_background_points)
ax.scatter(df_ss[:, 0], df_ss[:, 1], c="red", s=size_points)
if density:
cols = gaussian_kde(df_ss[:, [0, 1]].T)(df_ss[:, [0, 1]].T)
ax.scatter(df_ss[:, 0], df_ss[:, 1], c=cols, s=size_points)
if centers:
ax.scatter(df_c[:, 0], df_c[:, 1], c="black", s=10)
ax.set(xlabel=xy_label[0], ylabel=xy_label[1])
if x_lim is not None:
ax.set_xlim(x_lim)
if y_lim is not None:
ax.set_ylim(y_lim)
ax.set_title(title)
return fig


Expand Down
12 changes: 12 additions & 0 deletions tests/test_basic.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -43,6 +43,18 @@ def test_plot_pies(FlowSOM_res, gating_results):
pl.savefig("plotpies.pdf")


def test_plot_2D_scatters(FlowSOM_res):
pl = fs.pl.plot_2D_scatters(
FlowSOM_res,
channelpairs=[["CD3", "CD4"], ["CD19", "TCRb"]],
clusters=[[1, 2], [3]],
metaclusters=[[4], [5, 6]],
density=False,
centers=True,
)
pl.savefig("plot2Dscatters.pdf")


def test_new_data(fcs):
fsom = fs.FlowSOM(fcs[0:5000, :], cols_to_use=[8, 11, 13, 14, 15, 16, 17])
fsom_new = fsom.new_data(fcs[1001:2000])
Expand Down

0 comments on commit a53f4a6

Please sign in to comment.