Merge pull request #510 from bigbio/dev

Heatmap for FragPipe

Author: ypriverol

pmultiqc/modules/common/dia_utils.py

@@ -8,7 +8,10 @@
 from multiqc.plots import table
 
 from pmultiqc.modules.common.histogram import Histogram
-from pmultiqc.modules.common.stats import qual_uniform
+from pmultiqc.modules.common.stats import (
+    qual_uniform,
+    cal_hm_charge
+)
 from pmultiqc.modules.common.plots import dia as dia_plots
 from pmultiqc.modules.common.file_utils import file_prefix
 from pmultiqc.modules.common.common_utils import (
@@ -576,14 +579,11 @@ def heatmap_cont_pep_intensity(report_df):
     df["is_contaminant"] = df["Protein.Names"].str.startswith("CON", na=False)
 
     # 3. "Charge"
-    charge = dict()
-    for raw_file, group in df[["Run", "Precursor.Charge"]].groupby("Run"):
-        vc = group["Precursor.Charge"].value_counts(dropna=True, normalize=True)
-        charge[raw_file] = vc.get(2, 0)
-    charge_median = np.median(list(charge.values())) if charge else 0
-    heatmap_charge = {
-        k: float(max(0.0, 1.0 - abs(v - charge_median))) for k, v in charge.items()
-    }
+    heatmap_charge = cal_hm_charge(
+        df=df,
+        run_col="Run",
+        charge_col="Precursor.Charge"
+    )
 
     heatmap_dict = {}
     for run, group in df.groupby("Run"):

pmultiqc/modules/common/plots/general.py

@@ -95,7 +95,7 @@ def draw_heatmap(
     heatmap_data,
     heatmap_xnames,
     heatmap_ynames,
-    is_maxquant
+    report_type
 ):
     pconfig = {
         "id": "heatmap",
@@ -112,9 +112,12 @@ def draw_heatmap(
         "cluster_cols": False,
         "save_data_file": False,
     }
-    if is_maxquant:
+    if report_type == "maxquant":
         hm_html = heatmap.plot(data=heatmap_data, pconfig=pconfig)
         description_text = "This heatmap provides an overview of the performance of MaxQuant."
+    elif report_type == "fragpipe":
+        hm_html = heatmap.plot(data=heatmap_data, pconfig=pconfig)
+        description_text = "This heatmap provides an overview of the performance of FragPipe."
     else:
         hm_html = heatmap.plot(heatmap_data, heatmap_xnames, heatmap_ynames, pconfig)
         description_text = "This heatmap provides an overview of the performance of quantms."

pmultiqc/modules/common/stats.py

@@ -1,4 +1,5 @@
 import numpy as np
+import pandas as pd
 
 
 def nanmedian(values: np.ndarray, all_nan_fallback: np.float64) -> np.float64:
@@ -56,4 +57,22 @@ def cal_delta_mass_dict(df, col):
         "frequency": frequency_bin_data,
     }
 
-    return delta_mass
+    return delta_mass
+
+
+def cal_hm_charge(df: pd.DataFrame, run_col: str, charge_col: str):
+
+    if run_col not in df.columns or charge_col not in df.columns:
+        return {}
+
+    charge = dict()
+
+    for raw_file, group in df[[run_col, charge_col]].groupby(run_col):
+        vc = group[charge_col].value_counts(dropna=True, normalize=True)
+        charge[raw_file] = vc.get(2, 0)
+    charge_median = np.median(list(charge.values())) if charge else 0
+
+    return {
+        k: float(max(0.0, 1.0 - abs(v - charge_median)))
+        for k, v in charge.items()
+    }