rwe 0.0.1__py3-none-any.whl → 0.0.3__py3-none-any.whl

rwe/clients/aou.py

@@ -45,9 +45,13 @@ def generate_aou_variant_info_demographics_report(doc: Document, chrm: str, gene
 
 
 ############### Clinical Records ###############
-def clean_aou_phewas(phewas_file):
-    df = pd.read_csv(phewas_file, sep="\t")
-    df = df.loc[(df.ancestry == "all")&(df.converged==True)]
+def clean_aou_phewas(phewas_file, version="1.2"):
+    if version == "1.2":
+        df = pd.read_csv(phewas_file)
+        df = df.loc[df.converged==True]
+    elif version =="X":
+        df = pd.read_csv(phewas_file, sep="\t")
+        df = df.loc[(df.ancestry == "all")&(df.converged==True)]
     return df
 
 def get_aou_manhattan(df, gene):
@@ -70,7 +74,7 @@ def get_aou_manhattan(df, gene):
 
 def generate_aou_clinical_report(doc, chrm, gene, zygosity):
     from rwe.parsers.aou.config import BUCKET
-    phewas_file = f"{BUCKET}/data/phewas/results/{chrm}/{gene}_phewas.csv"
+    phewas_file = f"{BUCKET}/data/phewas/results/chr{chrm}/{gene}_phewas.csv"
     if uth._gcs_size(phewas_file, BUCKET) > 0: 
         df = clean_aou_phewas(phewas_file)
         fig, plot_df = get_aou_manhattan(df, gene)
@@ -107,29 +111,10 @@ def generate_aou_clinical_report(doc, chrm, gene, zygosity):
 
 
 ############### Labs and Measurements ###############
-def remove_outliers_iqr(df, column, multiplier=5):
-    """
-    Remove outliers from a specified column in a pandas DataFrame using the IQR method.
-    """
-    value = np.log1p(df[column].astype(float))
-    Q1 = value.quantile(0.25)
-    Q3 = value.quantile(0.75)
-    IQR = Q3 - Q1
-    lower_bound = Q1 - multiplier * IQR
-    upper_bound = Q3 + multiplier * IQR
-    return np.where(value.between(lower_bound, upper_bound), df[column], pd.NA)
-
-def clean_measurement(df, col="median_value", multiplier=5):
-    df =  df.copy()
-    df = df[df[col] >= 0]
-    df[col] = remove_outliers_iqr(df, col, multiplier)
-    return df.dropna(subset=[col])
-
 def compare(df, measurement):
     mdf = df.loc[df.measurement==measurement].copy()
     ctrls = mdf.loc[mdf.cases==False].drop_duplicates(["person_id", "measurement_concept_id"])
     cases = mdf.loc[mdf.cases==True].drop_duplicates(["person_id", "measurement_concept_id"])
-    ctrls = clean_measurement(ctrls)
     x = pd.to_numeric(cases["median_value"], errors="coerce").dropna().to_numpy()
     y = pd.to_numeric(ctrls["median_value"], errors="coerce").dropna().to_numpy()
     caq1, caq2, caq3 = pd.Series(x).quantile([0.25, 0.5, 0.75])
@@ -177,7 +162,7 @@ def generate_aou_labs_measurements_report(doc: Document, chrm: str, gene: str, z
     res_df = run_parallel(numerical_measurements_df, measurements=measurements, n_jobs=None, chunksize=20)
     # TODO: add most significant measurements table to doc
     for k,v in MEASUREMENT_GROUPS.items():
-        f,a,p = plot_measurements(numerical_measurements_df, v, multiplier=5, col="median_value", res_df=res_df)
+        f, a = plot_measurements(numerical_measurements_df, v, col="median_value", res_df=res_df)
         fig_path = uth._save_fig_to_tmp(f, basename=f"aou_measurements_{k}", dpi=300)
         doc.add_paragraph()  # spacing
         doc.add_picture(fig_path, width=Inches(6.5))
@@ -188,8 +173,8 @@ def generate_aou_labs_measurements_report(doc: Document, chrm: str, gene: str, z
 ############### Surveys ###############
 def clean_aou_surveys(df, survey_col="survey", question_col="question", answer_col="answer_category", zygosity="hetz"):
     df = df.copy()
-    df[question_col] = df[question_col].str.replace("^" + df[survey_col] + ": ", "", regex=True)
-    df[answer_col] = df[answer_col].str.replace("^" + df[question_col] + ": ", "", regex=True)
+    df[question_col] = df[question_col].str.replace(r"^.*?:\s*", "", regex=True, n=1)
+    df[answer_col] = df[answer_col].str.replace(r"^.*?:\s*", "", regex=True, n=1)
     return df
 
 def generate_aou_survey_report(doc: Document, chrm: str, gene: str, zygosity: str) -> Document:

rwe/parsers/aou/config.py

@@ -61,7 +61,7 @@ PLAUSIBLE = {
 
     # --- Liver / protein balance (AoU typical units: albumin g/L, protein g/L, enzymes U/L, bilirubin mg/dL OR umol/L; range is loose) ---
     "Albumin [Mass/volume] in Serum or Plasma": (15, 60),                   # g/L
-    "Protein [Mass/volume] in Serum or Plasma": (40, 100),                  # g/L
+    "Protein [Mass/volume] in Serum or Plasma": (3, 12),                  # g/dL
     "Alanine aminotransferase [Enzymatic activity/volume] in Serum or Plasma": (0, 1000),     # U/L
     "Aspartate aminotransferase [Enzymatic activity/volume] in Serum or Plasma": (0, 1000),   # U/L
     "Alkaline phosphatase [Enzymatic activity/volume] in Serum or Plasma": (10, 2000),        # U/L

rwe/parsers/aou/process.py

@@ -7,9 +7,29 @@ from scipy import stats
 import multiprocessing as mp
 from scipy.stats import ks_2samp
 
+def remove_outliers_iqr(df, column, multiplier=5):
+    """
+    Remove outliers from a specified column in a pandas DataFrame using the IQR method.
+    """
+    value = np.log1p(df[column].astype(float))
+    Q1 = value.quantile(0.25)
+    Q3 = value.quantile(0.75)
+    IQR = Q3 - Q1
+    lower_bound = Q1 - multiplier * IQR
+    upper_bound = Q3 + multiplier * IQR
+    return np.where(value.between(lower_bound, upper_bound), df[column], pd.NA)
+
+def remove_outliers(df, col="median_value", multiplier=5):
+    df =  df.copy()
+    df = df[df[col] >= 0]
+    df[col] = remove_outliers_iqr(df, col, multiplier)
+    return df.dropna(subset=[col])
 
 def clean_measurements_helper(g):
+    from rwe.parsers.aou.config import PLAUSIBLE, UNIT_DROPS, UNIT_CONVERSIONS
     m = g["measurement"].iat[0]
+    g["median_value"] = pd.to_numeric(g["median_value"], errors="coerce")
+    g = g.dropna(subset=["median_value"])
     # 1) drop units (manual)
     drops = UNIT_DROPS.get(m, set())
     if drops:
@@ -25,15 +45,17 @@ def clean_measurements_helper(g):
     # 3) plausible range filter
     lo, hi = PLAUSIBLE[m]
     g = g[g["median_value"].between(lo, hi)].copy()
-
-    # 4) remove units present in < MIN_UNIT_N samples
+    # 4) IQR outlier removal (pooled; before KS)
+    g = remove_outliers(g, col="median_value", multiplier=5)
+    if g.empty:
+        return g
+    # 5) remove units present in < MIN_UNIT_N samples
     unit_counts = g["unit"].value_counts()
     keep_units = unit_counts[unit_counts >= 5].index
     g = g[g["unit"].isin(keep_units)].copy()
     if g.empty or g["unit"].nunique() == 1:
         return g
-    
-    # 5) dissimilar distributions via KS test vs most common unit
+    # 6) dissimilar distributions via KS test vs most common unit
     ref_unit = g["unit"].value_counts().idxmax()
     ref = g.loc[g["unit"].eq(ref_unit), "median_value"].astype(float).dropna()
 
@@ -48,7 +70,7 @@ def clean_measurements_helper(g):
     return g[g["unit"].isin(keep)].copy()
 
 def clean_measurements():
-    from rwe.parsers.aou.config import BUCKET, CDR, GOOGLE_PROJECT
+    from rwe.parsers.aou.config import BUCKET, CDR, GOOGLE_PROJECT, PLAUSIBLE
     numerical_measurements_df = pd.read_parquet(f"{BUCKET}/data/rwe_info/raw/numerical_measurements.parquet")
     selected_nm_df = numerical_measurements_df.loc[
     numerical_measurements_df.measurement.isin(PLAUSIBLE.keys())
@@ -71,4 +93,3 @@ def clean_surveys():
     selected_survey_df = survey_df.loc[survey_df.question.isin(questions)].copy()
     selected_survey_df.to_parquet(f"{BUCKET}/data/rwe_info/processed/selected_surveys.parquet")
     return
-

rwe/plots/demographics.py

@@ -121,7 +121,7 @@ def demographics_plot(person_df,
                         top_n_sex=2,
                         top_n_ethnicity=2,
                         top_n_ancestry=None,
-                        figsize=(14, 7),
+                        figsize=(4, 5),
                         palette=None,
                         share_legend=True,
                         savepath=None,

rwe/plots/measurements.py

@@ -27,14 +27,26 @@ def fmt_p(p):
     return f"P={p:.2e}" if p < 0.001 else f"P={p:.3f}"
 
 def plot_measurements(
-    df_long, measurements, multiplier=5, col="median_value",  
-    res_df=None, res_name_col="measurement", res_p_col="p_mwu"
-):
+    df_long,
+    measurements,
+    col="median_value",
+    res_df=None,
+    res_name_col="measurement",
+    res_p_col="p_mwu",
+    ncols=3,
+    figsize=(6, 4),
+    palette=None
+    ):
     d = df_long[df_long["measurement"].isin(measurements)].copy()
     d[col] = pd.to_numeric(d[col], errors="coerce")
     d = d.dropna(subset=[col])
     d["group"] = np.where(d["cases"], "Cases", "Controls")
 
+    if palette is None:
+        palette = {
+            "Controls": "#2F6690",  # BLUE
+            "Cases":    "#D1495B",  # RED
+        }
     # star map
     p_map, star_map = {}, {}
     if res_df is not None:
@@ -43,25 +55,22 @@ def plot_measurements(
         p_map = dict(zip(tmp[res_name_col], tmp[res_p_col]))
         star_map = {k: p_to_stars(v) for k, v in p_map.items()}
         
-    parts = []
-    for m in measurements:
-        sub = d[d["measurement"] == m].copy()
-        sub = apply_plausible_range(sub, m)
-        ctrl = clean_measurement(sub[sub["group"] == "Controls"], col=col, multiplier=multiplier)
-        case = sub[sub["group"] == "Cases"].copy()  # don't IQR-clean cases
-        parts.append(pd.concat([ctrl, case], ignore_index=True))
-    p = pd.concat(parts, ignore_index=True)
-
-    fig, axes = plt.subplots(2, 3, figsize=(4, 3))
-    axes = axes.flatten()
+    # Layout: dynamic grid
+    n = len(measurements)
+    nrows = int(np.ceil(n / ncols))
+    fig, axes = plt.subplots(nrows, ncols, figsize=figsize)
+    axes = np.array(axes).ravel()
 
     for i, m in enumerate(measurements):
         ax = axes[i]
-        sub = p[p["measurement"] == m].copy()
-        sub[col] = pd.to_numeric(sub[col], errors="coerce")
-        sub = sub.dropna(subset=[col])
-        sns.violinplot(data=sub, x="group", y=col, order=["Controls", "Cases"],
-                       inner="quartile", cut=0, linewidth=1, ax=ax)
+        sub = d[d["measurement"] == m].copy()
+
+        sns.violinplot(
+            data=sub, x="group", y=col, order=["Controls", "Cases"],
+            hue="group", hue_order=["Controls", "Cases"], legend=False,
+            palette=palette, inner="quartile", cut=0, linewidth=1, ax=ax
+            )
+
         ax.set_title(textwrap.fill(m, width=20,  max_lines=3))
         ax.set_xlabel("")
         ax.set_ylabel("")
@@ -75,15 +84,20 @@ def plot_measurements(
             ymin = np.nanmin(sub[col].values)
             span = (ymax - ymin) if ymax > ymin else 1.0
 
-            y_line = ymax + 0.07 * span
-            y_star = y_line + 0.02 * span
-            y_p = y_star - 0.03 * span  # right below stars
-
+            # put annotation above violins
+            y_txt = ymax + 0.08 * span
+            
+            # x=0.5 is between the two categories (0 and 1)
+            if stars:
+                ax.text(0.5, y_txt + 0.03 * span, stars, ha="center", va="bottom")
             if not pd.isna(pval):
-                ax.text(0.5, y_p, fmt_p(pval), ha="center", va="top")
+                ax.text(0.5, y_txt, fmt_p(pval), ha="center", va="bottom")
+
+            # give headroom so text doesn't clip
+            ax.set_ylim(ymin, y_txt + 0.12 * span)
 
-    for j in range(len(vitals), 6):
+    for j in range(n, len(axes)):
         axes[j].axis("off")
 
     plt.tight_layout()
-    return fig, axes, p
+    return fig, axes

rwe/plots/surveys.py

@@ -84,7 +84,7 @@ def percent_bar_with_n(
         palette=palette
     )
 
-    ax.set_title(title or f"{cat_col} (% within group)")
+    ax.set_title(title or f"{cat_col} (% within group)", pad=20)
     ax.set_xlabel("")
     ax.set_ylabel("Percent (%)")
     ax.tick_params(axis="x", rotation=25)
@@ -158,6 +158,7 @@ def plot_survey_questions(
     
     fig, axes = plt.subplots(nrows, ncols, figsize=figsize)
     axes = axes.ravel()
+    handles = labels = None
 
     for i, ((svy, q), sub) in enumerate(groups):
         percent_bar_with_n(

rwe/plots/variant_info.py

@@ -48,7 +48,7 @@ def create_variant_frequency_plots(
     bar_color="#D1495B",
     bar_fontsize=6,
     table_fontsize=6,
-    bar_height_ratio=1.5,            # relative height of bar panel vs table panel
+    bar_height_ratio=0.25,            # relative height of bar panel vs table panel
 ):
     df = variant_df.copy()
 

{rwe-0.0.1.dist-info → rwe-0.0.3.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: rwe
-Version: 0.0.1
+Version: 0.0.3
 Summary: Real World Evidence utilities and reporting
 Author: Deepro Banerjee
 License: MIT License
@@ -32,7 +32,7 @@ License-File: LICENSE
 Requires-Dist: pandas
 Requires-Dist: numpy
 Requires-Dist: matplotlib
-Requires-Dist: seaborn>=0.12
+Requires-Dist: seaborn>=0.13
 Requires-Dist: python-docx>=1.1.0
 Requires-Dist: tqdm
 Requires-Dist: requests

{rwe-0.0.1.dist-info → rwe-0.0.3.dist-info}/RECORD

@@ -2,26 +2,26 @@ rwe/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 rwe/generate_report.py,sha256=-NbQJ4H-NYdDzziWMoi5PUWyhZ7p8iVC8UXYUP1v_xY,1833
 rwe/assets/Arrowhead_Pharmaceuticals_logo.png,sha256=C2mAn6GG4gArds4sBnqM5LcxhuZgujXjwGiZxIuxbLo,51053
 rwe/clients/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-rwe/clients/aou.py,sha256=7nPg2JU1LtkQ5AClqQMpeqjWh3kBrI63RmYOFcfVqxc,9949
+rwe/clients/aou.py,sha256=XVvRArzcTTBO0JmiTtWvuiafTuzpupaZJBb1Mtt8QIQ,9387
 rwe/clients/azn.py,sha256=6zwEv0IAgBA-FcI2glGerA9M6pzdKQx7M5M4HZI8HQs,7541
 rwe/clients/genebass.py,sha256=OkF0_J4lmiujWlffsBlAM3uKxKOCzi7_3f_ehju_tPE,6103
 rwe/clients/hgnc.py,sha256=ZabNiMgzwu4rCVCptbgUb38qHsL9cHlsQX7s6mZA8Nk,2015
 rwe/parsers/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 rwe/parsers/aou/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 rwe/parsers/aou/collect.py,sha256=3j9S0vg3s5u-lJYhP9gdJA_SJVUTUld-i8M1mS-qg-0,13610
-rwe/parsers/aou/config.py,sha256=_BdQohVgSeLNIkoLBQ4XIgo2ucep6NDU8k0vGKgONj8,27018
-rwe/parsers/aou/process.py,sha256=T3B4Ir3qHoq7txEO6Bxi1L_qHqjTbWikVLtFHS6QuSM,2705
+rwe/parsers/aou/config.py,sha256=7O8NhrxOAeDJnKYsiQBgufNkxo_O6nRPSrhJ49pteLk,27017
+rwe/parsers/aou/process.py,sha256=q-H12OQdwBZMqWxuDo389m6WCJb9Qsp5ZIwVc2MHqu4,3683
 rwe/plots/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 rwe/plots/clinical.py,sha256=dBgdrFog-4gWJyJphoObSuoPtS7Q4wOprWyXMcDD6Dw,7738
-rwe/plots/demographics.py,sha256=QrDA1QSKG56Iuv9lodGfYBTymBOf0WKm0q6QFzs0SB4,7288
-rwe/plots/measurements.py,sha256=gKSWcon9Wzb3_oO4CnBRn3GTbNqjkfILXj6CmY3josQ,3177
-rwe/plots/surveys.py,sha256=X8DT2vnnA1S2u98mrzcA-apSOht8BAsCFjMbHQj7pfY,6636
-rwe/plots/variant_info.py,sha256=7V9dJWc8XEQU_t7KY1NzldxYbfOPMKRYBiSNIysizB4,5361
+rwe/plots/demographics.py,sha256=TqYY0dxroLAq9YOZHo1-SebacvNZpQjzUSui-nBUKSg,7287
+rwe/plots/measurements.py,sha256=Z7GG4SMXD3ExGGJF6LvmZ_XwvMwGB_6FFk7nxz_8fr8,3286
+rwe/plots/surveys.py,sha256=2SiNBnjLQL2V4jfeMTIj2y2v7y7Wh4kLji2dnEJ0cdo,6672
+rwe/plots/variant_info.py,sha256=KLUnefwmSxcmOn7mkNBcxvJdpNN4i4F9SZ0SZGlNrvk,5362
 rwe/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 rwe/utils/helpers.py,sha256=t3fmx2OsFx3fvg2uZggagm5LS-PE7J3Yr7Ew3VJUsnA,4336
 rwe/utils/report.py,sha256=Vyfle1aOoCkqtVeBHAzCdb6hgR0sQZn_qQ4szUw_A3w,4658
-rwe-0.0.1.dist-info/licenses/LICENSE,sha256=0PfJPAoyFVWY4L80aEcQyjy4PojxzfqcOenSuXdhgBg,1072
-rwe-0.0.1.dist-info/METADATA,sha256=x5Fi7T0sZPjz3wiZyhB8oyS_VU6O5IRYJ4RmOqPyV1w,23601
-rwe-0.0.1.dist-info/WHEEL,sha256=wUyA8OaulRlbfwMtmQsvNngGrxQHAvkKcvRmdizlJi0,92
-rwe-0.0.1.dist-info/top_level.txt,sha256=zwX0M3dfnEfklZqvvBPQXiJykChpiqt219EolC3WFRc,4
-rwe-0.0.1.dist-info/RECORD,,
+rwe-0.0.3.dist-info/licenses/LICENSE,sha256=0PfJPAoyFVWY4L80aEcQyjy4PojxzfqcOenSuXdhgBg,1072
+rwe-0.0.3.dist-info/METADATA,sha256=huTbgeLgndyhlIZHX4qOEPQi-jNFCpfRUPF2FM__EZI,23601
+rwe-0.0.3.dist-info/WHEEL,sha256=wUyA8OaulRlbfwMtmQsvNngGrxQHAvkKcvRmdizlJi0,92
+rwe-0.0.3.dist-info/top_level.txt,sha256=zwX0M3dfnEfklZqvvBPQXiJykChpiqt219EolC3WFRc,4
+rwe-0.0.3.dist-info/RECORD,,