beta_mds_plot requires at least 2 samples now, to avoid MDS warnings

methylcheck/qc_report.py

@@ -303,7 +303,6 @@ def __init__(self, **kwargs):
         # https://stackoverflow.com/questions/8187082/how-can-you-set-class-attributes-from-variable-arguments-kwargs-in-python
         self.__dict__.update(kwargs)
         self.debug = True if self.__dict__.get('debug') == True else False
-        self.__dict__.pop('debug',None)
         self.__dict__['poobah_max_percent'] = self.__dict__.get('poobah_max_percent', 5)
         self.__dict__['pval_cutoff'] = self.__dict__.get('pval_cutoff', 0.05)
         self.errors = self.open_error_buffer()
@@ -443,7 +442,7 @@ def run_qc(self):
                         self.to_table(list_of_lists, col_names=['Sample_ID', 'Percent', 'Pass/Fail'],
                             row_names=None, add_title='Detection Poobah')
 
-                    if part == 'mds':
+                    if part == 'mds' and len(beta_df.columns) > 1:
                         LOGGER.info("Beta MDS Plot")
                         # ax and df_to_retain are not used, but could go into a qc chart
                         fig, ax, df_indexes_to_retain = methylcheck.beta_mds_plot(beta_df, silent=True, multi_params={'return_plot_obj':True, 'draw_box':True})

methylcheck/samples/postprocessQC.py

@@ -363,6 +363,9 @@ def beta_mds_plot(df, filter_stdev=1.5, verbose=False, save=False, silent=False,
     if verbose:
         logging.basicConfig(level=logging.INFO)
 
+    if len(df.columns) < 2:
+        LOGGER.warning("beta_mds_plot requires at least 2 samples")
+        return df
 
     # ensure "long format": probes in rows and samples in cols. This is how methylprep returns data.
     if df.shape[1] < df.shape[0]:
@@ -372,6 +375,7 @@ def beta_mds_plot(df, filter_stdev=1.5, verbose=False, save=False, silent=False,
         if verbose:
             LOGGER.info(f"Your data needed to be transposed (from {pre_df_shape} to {df.shape}) to ensure probes are in columns.")
     original_df = df.copy() # samples in index, guaranteed. transpose at end
+    # require 2 or more samples for MDS
 
     # CHECK for missing probe values NaN -- this is common as of methylprep version 1.2.5 because pOOBah removes probes from samples by default.
     missing_probe_counts = df.isna().sum()

tests/reports/test_qc_report.py

@@ -25,13 +25,26 @@ def test_qc_run_pipeline(self):
         df = methylcheck.load(PROCESSED_450K)
         methylcheck.run_pipeline(df, exclude_all=True, plot=['all'], silent=True)
 
+
     def test_ReportPDF(self):
         import warnings
-        warnings.filterwarnings('ignore')
+        warnings.filterwarnings("ignore", message='invalid value encountered')
         myreport = methylcheck.ReportPDF(path=PROCESSED_450K, outpath=PROCESSED_450K)
         myreport.run_qc()
         myreport.pdf.close()
         if Path(PROCESSED_450K,'multipage_pdf.pdf').exists():
             Path(PROCESSED_450K,'multipage_pdf.pdf').unlink()
         else:
             raise FileNotFoundError(Path(PROCESSED_450K,'multipage_pdf.pdf'))
+
+
+    def test_dummy(self):
+        import warnings
+        from sklearn.manifold import MDS
+        warnings.filterwarnings("ignore", message='invalid value encountered')
+        warnings.filterwarnings("ignore", category=RuntimeWarning)
+        #df = methylcheck.load(PROCESSED_450K)
+        df = pd.DataFrame(data={'9247377093_R02C01':[0.1,.2,.3,.4,.5,.6,np.nan,np.nan,0.8,0.9,1.0],
+        'two':[0.1,.2,.3,.4,.5,np.nan,.6,np.nan,0.8,0.9,0.1]})
+        mds = MDS(n_jobs=-1, random_state=1, verbose=1)
+        mds_transformed = mds.fit_transform(df.dropna().transpose().values)