Benchmark GRN inference

Benchmark GRN inference#

Library imports#

import pandas as pd

import matplotlib.pyplot as plt
import mplscience
import seaborn as sns

from rgv_tools import DATA_DIR, FIG_DIR

Constants#

DATASET = "mHSPC"

SAVE_DATA = True
if SAVE_DATA:
    (DATA_DIR / DATASET / "results").mkdir(parents=True, exist_ok=True)

SAVE_FIGURE = True
if SAVE_FIGURE:
    (FIG_DIR / DATASET).mkdir(parents=True, exist_ok=True)

Data loading#

tfv_score = pd.read_csv(DATA_DIR / DATASET / "results" / "GRN_benchmark_tfv.csv")
rgv_score = pd.read_csv(DATA_DIR / DATASET / "results" / "GRN_benchmark_rgv.csv")
grn_score = pd.read_csv(DATA_DIR / DATASET / "results" / "GRN_benchmark.csv")

df = pd.concat([tfv_score, rgv_score, grn_score], ignore_index=True)

Plot the benchmark#

df
Unnamed: 0 EPR AUC Method
0 0 1.116878 0.639869 tfvelo
1 0 1.138696 0.603134 regvelo
2 1 1.145427 0.600968 regvelo
3 2 1.124305 0.594723 regvelo
4 0 1.108522 0.566040 Corr
5 1 1.014753 0.533928 GRNBoost2
6 2 1.196337 0.557187 CellOracle 
with mplscience.style_context():
    sns.set_style(style="whitegrid")
    fig, ax = plt.subplots(figsize=(3, 3), sharey=True)
    # Plot the second Seaborn plot on the first subplot
    sns.barplot(
        y="Method",
        x="AUC",
        data=df,
        capsize=0.1,
        color="grey",
        order=["regvelo", "Corr", "CellOracle", "GRNBoost2", "tfvelo"],
        ax=ax,
    )
    sns.stripplot(
        y="Method",
        x="AUC",
        data=df,
        order=["regvelo", "Corr", "CellOracle", "GRNBoost2", "tfvelo"],
        color="black",
        size=3,
        jitter=True,
        ax=ax,
    )
    ax.set_xlabel("AUC", fontsize=14)
    ax.set_ylabel("")
    plt.xlim(0.5, 0.7)
    plt.show()

    if SAVE_FIGURE:
        fig.savefig(FIG_DIR / DATASET / "AUC.svg", format="svg", transparent=True, bbox_inches="tight")
../_images/d0a774a8065ac3409c83a694b38d6628f3ff4e5e8156f826f0b13da8e5f92c03.png 
with mplscience.style_context():
    sns.set_style(style="whitegrid")
    fig, ax = plt.subplots(figsize=(3, 3), sharey=True)
    # Plot the second Seaborn plot on the first subplot
    sns.barplot(
        y="Method",
        x="EPR",
        data=df,
        capsize=0.1,
        color="grey",
        order=["regvelo", "Corr", "CellOracle", "GRNBoost2", "tfvelo"],
        ax=ax,
    )
    sns.stripplot(
        y="Method",
        x="EPR",
        data=df,
        order=["regvelo", "Corr", "CellOracle", "GRNBoost2", "tfvelo"],
        color="black",
        size=3,
        jitter=True,
        ax=ax,
    )
    ax.set_xlabel("EPR", fontsize=14)
    ax.set_ylabel("")
    plt.xlim(
        1.0,
    )
    plt.show()

    if SAVE_FIGURE:
        fig.savefig(FIG_DIR / DATASET / "EPR.svg", format="svg", transparent=True, bbox_inches="tight")
../_images/971ee67c284abcca8dc5a66437a934ee393c76e857110def6b7daf0105bbe09c.png
Contents