switch to polars

otherwise interger columns with missing data are not read as ints

app.py

@@ -4,8 +4,7 @@ from typing import Callable, Optional, cast
 from urllib.parse import parse_qsl
 
 import altair as alt
-import numpy as np
-import pandas as pd
+import polars as pl
 import reacton.core
 import solara
 import solara.lab
@@ -91,7 +90,7 @@ def ColorPickerMenuButton(title: str, color: solara.Reactive[str]):
         )
 
 
-empty_frame = pd.DataFrame()
+empty_frame = pl.DataFrame()
 R_DEFAULT = ""
 V_DEFAULT = ""
 
@@ -130,17 +129,17 @@ def MainApp():
 
     def on_file(file_info: solara.components.file_drop.FileInfo | None):
         if not file_info:
-            data.set(pd.DataFrame())
+            data.set(pl.DataFrame())
             return
 
         try:
-            df = pd.read_csv(file_info["file_obj"])
+            df = pl.read_csv(file_info["file_obj"])
         except Exception as e:
             warning_text.set(str(e))
             return
         if len(df.columns) < 2:
             warning_text.set(f"Expected at least 2 columns, got {len(df.columns)}")
-            data.set(pd.DataFrame())
+            data.set(pl.DataFrame())
             return
 
         warning_text.set("")
@@ -168,10 +167,10 @@ def MainApp():
         autoscale_y=autoscale_y.value,
     )
 
-    if data.value.empty:
-        data_view = pd.DataFrame({"residue_number": [], "value": []})
+    if data.value.is_empty():
+        data_view = pl.DataFrame({"residue_number": [], "value": []})
     else:
-        data_view = pd.DataFrame(
+        data_view = pl.DataFrame(
             {
                 "residue_number": data.value[residue_column.value],
                 "value": data.value[color_column.value],
@@ -236,7 +235,7 @@ def MainApp():
             if warning_text.value:
                 solara.Warning(warning_text.value)
 
-            if not data.value.empty:
+            if not data.value.is_empty():
                 with solara.Row():
                     solara.Select(
                         label="Residue Column",
@@ -258,10 +257,17 @@ def MainApp():
                 ColorPickerMenuButton("Highlight", highlight_color)
                 ColorPickerMenuButton("Missing data", missing_data_color)
 
+            def set_cmap_name(name: str):
+                try:
+                    Colormap(name)
+                    cmap_name.set(name)
+                except TypeError:
+                    pass
+
             # with solara.Row():
             solara.v.Autocomplete(
                 v_model=cmap_name.value,
-                on_v_model=cmap_name.set,
+                on_v_model=set_cmap_name,
                 items=CMAP_OPTIONS,
             )
 

make_link.py

@@ -5,7 +5,7 @@ from io import BytesIO
 from typing import TYPE_CHECKING
 from urllib.parse import urlencode
 
-import pandas as pd
+import polars as pl
 import zstandard as zstd
 
 if TYPE_CHECKING:
@@ -20,13 +20,13 @@ def encode_url(
     molecule_id: str,
     colors: ColorTransform,
     axis_properties: AxisProperties,
-    data: pd.DataFrame,
+    data: pl.DataFrame,
     description: str = "",
 ):
     encode_dict = dict(title=title, molecule_id=molecule_id)
     encode_dict.update({**colors.model_dump(), **axis_properties.model_dump()})
 
-    csv_str = data.to_csv(float_format="%.4f", index=False)
+    csv_str = data.write_csv(float_precision=4, float_scientific=True)
     compressed = COMPRESSOR.compress(csv_str.encode())
     base64_text = base64.b64encode(compressed).decode("utf8")
     encode_dict["data"] = base64_text
@@ -36,12 +36,12 @@ def encode_url(
     return urlencode(encode_dict)
 
 
-def decode_data(base64_text) -> pd.DataFrame:
+def decode_data(base64_text) -> pl.DataFrame:
     decoded_bytes = base64.b64decode(base64_text)
     decompressed = DECOMPRESSOR.decompress(decoded_bytes)
 
     bio = BytesIO(decompressed)
-    data = pd.read_csv(bio)
+    data = pl.read_csv(bio)
     bio.close()
 
     return data

viewer.py

@@ -11,7 +11,7 @@ from urllib.parse import parse_qsl
 import altair as alt
 import ipywidgets as widgets
 import numpy as np
-import pandas as pd
+import polars as pl
 import solara
 import solara.lab
 from cmap import Colormap
@@ -36,7 +36,8 @@ class ColorTransform(BaseModel):
     missing_data_color: str = "#8c8c8c"
     highlight_color: str = "#e933f8"
 
-    def molstar_colors(self, data: pd.DataFrame) -> dict:
+    def molstar_colors(self, data: pl.DataFrame) -> dict:
+        data = data.drop_nulls()
         if self.norm_type == "categorical":
             values = data["value"]
         else:
@@ -85,7 +86,7 @@ class AxisProperties(BaseModel):
 
 
 def make_chart(
-    data: pd.DataFrame, colors: ColorTransform, axis_properties: AxisProperties
+    data: pl.DataFrame, colors: ColorTransform, axis_properties: AxisProperties
 ) -> alt.LayerChart:
     xmin, xmax = data["residue_number"].min(), data["residue_number"].max()
     xpad = (xmax - xmin) * 0.05
@@ -163,7 +164,7 @@ def make_chart(
 
     line_position = alt.param(name="line_position", value=0.0)
     line_opacity = alt.param(name="line_opacity", value=1)
-    df_line = pd.DataFrame({"x": [1.0]})
+    df_line = pl.DataFrame({"x": [1.0]})
 
     # Create vertical rule with parameter
     vline = (
@@ -188,7 +189,7 @@ def make_chart(
 
 @solara.component
 def ScatterChart(
-    data: pd.DataFrame,
+    data: pl.DataFrame,
     colors: ColorTransform,
     axis_properties: AxisProperties,
     on_selections,
@@ -223,11 +224,17 @@ def ScatterChart(
     solara.use_effect(bind, [data, colors])
 
 
+def is_numeric(val) -> bool:
+    if val is not None:
+        return not np.isnan(val)
+    return False
+
+
 @solara.component
 def ProteinView(
     title: str,
     molecule_id: str,
-    data: pd.DataFrame,
+    data: pl.DataFrame,
     colors: ColorTransform,
     axis_properties: AxisProperties,
     dark_effective: bool,
@@ -242,16 +249,17 @@ def ProteinView(
     # residue number to highlight in protein view
     highlight_number = solara.use_reactive(None)
 
-    if data.empty:
+    if data.is_empty():
         color_data = {}
     else:
         color_data = colors.molstar_colors(data)
+
     tooltips = {
         "data": [
             {
                 "residue_number": resi,
                 "tooltip": f"{axis_properties.label}: {value:.2g} {axis_properties.unit}"
-                if not np.isnan(value)
+                if is_numeric(value)
                 else "No data",
             }
             for resi, value in zip(data["residue_number"], data["value"])
@@ -317,7 +325,7 @@ def ProteinView(
             ).key(f"molstar-{dark_effective}")
         if not fullscreen.value:
             with solara.Card(style={"height": "550px"}):
-                if data.empty:
+                if data.is_empty():
                     solara.Text("No data")
                 else:
                     ScatterChart(
@@ -350,39 +358,3 @@ def RoutedView():
         )
     except KeyError as err:
         solara.Warning(f"Error: {err}")
-
-
-@solara.component
-def Page():
-    dark_effective = solara.lab.use_dark_effective()
-    dark_previous = solara.use_previous(dark_effective)
-
-    if dark_previous != dark_effective:
-        if dark_effective:
-            alt.themes.enable("dark")
-        else:
-            alt.themes.enable("default")
-
-    solara.Style(
-        """
-        .vega-embed {
-        overflow: visible;
-        width: 100% !important;
-        }"""
-    )
-
-    settings = json.loads(Path("settings.json").read_text())
-
-    colors = ColorTransform(**settings)
-    axis_properties = AxisProperties(**settings)
-
-    data = pd.read_csv("color_data.csv")
-
-    ProteinView(
-        settings["title"],
-        molecule_id=settings["molecule_id"],
-        data=data,
-        colors=colors,
-        axis_properties=axis_properties,
-        dark_effective=dark_effective,
-    )