Minor modifications to prefit

navicat_spock/helpers.py

@@ -142,6 +142,8 @@ def augment(df, level, verb=0):
         if verb > 0:
             print(f"Doing level 1 feature augmentation...")
         for i in x_full.keys():
+            if "TARGET" in i.upper():
+                continue
             inv = f"1/{i}"
             pow2 = f"{i}^2"
             # pow3 = f"{i}^3"
@@ -159,6 +161,8 @@ def augment(df, level, verb=0):
         if verb > 0:
             print(f"Doing level 2 feature augmentation...")
         for i, j in zip(x_full.keys(), x_full.keys()[1:]):
+            if "TARGET" in i.upper() or "TARGET" in j.upper():
+                continue
             prod = f"{i}x{j}"
             div12 = f"{i}/{j}"
             div21 = f"{j}/{i}"
@@ -176,6 +180,8 @@ def augment(df, level, verb=0):
                 continue
             i = x_full.keys()[i]
             j = x_full.keys()[j]
+            if "TARGET" in i.upper() or "TARGET" in j.upper():
+                continue
             prod = f"{i}x{j}"
             div12 = f"{i}/{j}"
             div21 = f"{j}/{i}"
@@ -225,7 +231,9 @@ def __exit__(self, *args):
 
 def namefixer(filename):
     assert isinstance(filename, str)
-    return re.sub("[^a-zA-Z0-9 \n\.]", "_", filename).replace(" ", "_")
+    return (
+        re.sub("[^a-zA-Z0-9 \n\.]", "_", filename).replace(" ", "_").replace("__", "_")
+    )
 
 
 def normalize(a, axis=-1, order=2):
@@ -367,7 +375,7 @@ def processargs(arguments):
         epilog="Remember to cite the spock paper (when its out!) \n \n - and enjoy!",
     )
     vbuilder.add_argument(
-        "-version", "--version", action="version", version="%(prog)s 0.0.4"
+        "-version", "--version", action="version", version="%(prog)s 0.0.5"
     )
     vbuilder.add_argument(
         "-i",

navicat_spock/plotting2d.py

@@ -80,6 +80,7 @@ def plot_2d(
     breakpoints=None,
     estimates=None,
     plotmode=1,
+    return_value=True,
     save_fig=True,
 ):
     # Labels and key
@@ -132,7 +133,11 @@ def plot_2d(
         plt.savefig(filename)
     if os.name != "posix" and "DISPLAY" in os.environ:
         plt.show()
-    return fig, ax
+    if return_value:
+        return fig, ax
+    else:
+        plt.close()
+        return None, None
 
 
 def plot_and_save(
@@ -193,6 +198,7 @@ def plot_and_save(
         estimates=estimates,
         plotmode=plotmode,
         filename=f"{namefixer(tags[idx].strip())}_volcano.png",
+        return_value=return_value,
         save_fig=save_fig,
     )
 
@@ -204,10 +210,13 @@ def plot_and_save(
         zdata = list(zip(xint, yint))
         csvname = f"{namefixer(tags[idx].strip())}_volcano.csv"
         np.savetxt(
-            csvname, zdata, fmt="%.4e", delimiter=",", header="{tags[idx]},{tags[tidx]}"
+            csvname,
+            zdata,
+            fmt="%.4e",
+            delimiter=",",
+            header="f{tags[idx]},{tags[tidx]}",
         )
     if return_value:
         return fig, ax
     else:
-        plt.close()
-        return None
+        return None, None

navicat_spock/spock.py

@@ -75,6 +75,10 @@ def run_spock_from_args(
         fig, ax = plt.subplots(
             frameon=False, figsize=[4.2, 3], dpi=300, constrained_layout=True
         )
+    if prefit:
+        fig_r, ax_r = plt.subplots(
+            frameon=False, figsize=[4.2, 3], dpi=300, constrained_layout=True
+        )
 
     # Atttempts to group data points based on shared characters in names.
     if setcbms:
@@ -211,16 +215,25 @@ def run_spock_from_args(
                     print(
                         f"Prefitting volcano with {n} breakpoints and descriptor index {idx}: {tags[idx]}, for which a BIC of {min_bic} was obtained."
                     )
-                _, _ = plot_and_save(
+                if fig_r is None and ax_r is None:
+                    fig_r, ax_r = plt.subplots(
+                        frameon=False,
+                        figsize=[4.2, 3],
+                        dpi=300,
+                        constrained_layout=True,
+                    )
+                fig_r, ax_r = plot_and_save(
                     pw_fit,
                     tags,
                     idx,
                     tidx,
                     cb,
                     ms,
-                    plotmode=plotmode,
                     fig=fig,
                     ax=ax,
+                    plotmode=plotmode,
+                    save_fig=save_fig,
+                    save_csv=save_csv,
                     return_value=False,
                 )
 
@@ -318,6 +331,13 @@ def run_spock_from_args(
                 pw_fit.best_muggeo = msels[idx - 1].models[n - 1].best_muggeo
             if verb > 2:
                 pw_fit.summary()
+            if fig is None and ax is None:
+                fig, ax = plt.subplots(
+                    frameon=False,
+                    figsize=[4.2, 3],
+                    dpi=300,
+                    constrained_layout=True,
+                )
             # Plot the data, fit, breakpoints and confidence intervals
             fig, ax = plot_and_save(
                 pw_fit,
@@ -331,6 +351,7 @@ def run_spock_from_args(
                 plotmode=plotmode,
                 save_fig=save_fig,
                 save_csv=save_csv,
+                return_value=True,
             )
             return fig, ax
         else:

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "navicat-spock"
-version = "0.0.4"
+version = "0.0.5"
 authors = [
   { name="R. Laplaza", email="rlaplaza@duck.com" },
 ]

setup.py

@@ -12,7 +12,7 @@
 setup(
     name="spock",
     packages=["navicat_spock"],
-    version="0.0.4",
+    version="0.0.5",
     description="Volcano Plot fitting tool",
     long_description=long_description,
     long_description_content_type="text/markdown",