Sequence plot

.gitignore

@@ -8,4 +8,4 @@ dist/
 docs/build
 docs/gallery
 docs/generated
-docs/gallery_scripts
+docs/gallery_scripts

pyopenms_viz/_bokeh/__init__.py

@@ -8,6 +8,7 @@
     BOKEHLinePlot,
     BOKEHVLinePlot,
     BOKEHScatterPlot,
+    BOKEHSequencePlot,
     BOKEHChromatogramPlot,
     BOKEHMobilogramPlot,
     BOKEHSpectrumPlot,
@@ -25,6 +26,7 @@
     "line": BOKEHLinePlot,
     "vline": BOKEHVLinePlot,
     "scatter": BOKEHScatterPlot,
+    "sequence": BOKEHSequencePlot,
     "chromatogram": BOKEHChromatogramPlot,
     "mobilogram": BOKEHMobilogramPlot,
     "spectrum": BOKEHSpectrumPlot,

pyopenms_viz/_bokeh/core.py

@@ -27,6 +27,7 @@
     LinePlot,
     VLinePlot,
     ScatterPlot,
+    SequencePlot,
     BaseMSPlot,
     ChromatogramPlot,
     MobilogramPlot,
@@ -412,6 +413,11 @@ def plot(cls, fig, data, x, y, by: str | None = None, plot_3d=False, **kwargs):
             return fig, legend
 
 
+class BOKEHSequencePlot(BOKEHPlot, SequencePlot):
+    def plot(self):
+        raise NotImplementedError("Sequence plot is not implemented for Plotly")
+
+
 class BOKEH_MSPlot(BaseMSPlot, BOKEHPlot, ABC):
 
     def get_line_renderer(self, data, x, y, **kwargs) -> None:

pyopenms_viz/_core.py

@@ -24,7 +24,7 @@
 from .constants import IS_SPHINX_BUILD
 
 
-_common_kinds = ("line", "vline", "scatter")
+_common_kinds = ("line", "vline", "scatter", "sequence")
 _msdata_kinds = ("chromatogram", "mobilogram", "spectrum", "peakmap")
 _all_kinds = _common_kinds + _msdata_kinds
 _entrypoint_backends = ("ms_matplotlib", "ms_bokeh", "ms_plotly")
@@ -446,6 +446,65 @@ def _kind(self):
         return "scatter"
 
 
+class SequencePlot(BasePlot, ABC):
+    """
+    Plot peptide sequence with matched fragments indicated.
+
+    Plot Specific Parameters
+    ------------------------
+    seq_col : string, optional
+        The name for sequence column
+    ion_annotation : string, optional
+        The name for the ion annotation column
+    color_annotation : string, optional
+        The name for the color annotation column
+    x_pos : float, optional
+        The center horizontal position of the peptide sequence.
+    y_pos : float, optional
+        The center vertical position of the peptide sequence.
+    spacing : float, optional
+        The horizontal spacing between amino acids.
+    seq_fontsize : str, optional
+        The font size of the amino acids.
+    frag_len : float, optional
+        The length of the fragment lines.
+    """
+
+    @property
+    def _kind(self):
+        return "sequence"
+
+    def __init__(
+        self,
+        data: DataFrame,
+        x,
+        y,
+        seq_col="sequence",
+        ion_annotation: str = "ion_annotation",
+        color_annotation: str = "color_annotation",
+        x_pos: float = 0.5,
+        y_pos: float = 0.5,
+        spacing: float = 0.06,
+        seq_fontsize: str | float = "xx-large",
+        frag_len: float = 0.06,
+        **kwargs,
+    ):
+        self.seq_col = seq_col
+        self.ion_annotation = ion_annotation
+        self.color_annotation = color_annotation
+        self.x_pos = x_pos
+        self.y_pos = y_pos
+        self.spacing = spacing
+        self.frag_len = frag_len
+        self.seq_fontsize = seq_fontsize
+
+        # Set default config attributes if not passed as keyword arguments
+        kwargs["_config"] = _BasePlotConfig(kind=self._kind)
+
+        super().__init__(data, x, y, **kwargs)
+        self.plot()
+
+
 class BaseMSPlot(BasePlot, ABC):
     """
     Abstract class for complex plots, such as chromatograms and mobilograms which are made up of simple plots such as ScatterPlots, VLines and LinePlots.

pyopenms_viz/_matplotlib/__init__.py

@@ -7,6 +7,7 @@
     MATPLOTLIBLinePlot,
     MATPLOTLIBVLinePlot,
     MATPLOTLIBScatterPlot,
+    MATPLOTLIBSequencePlot,
     MATPLOTLIBChromatogramPlot,
     MATPLOTLIBMobilogramPlot,
     MATPLOTLIBSpectrumPlot,
@@ -23,6 +24,7 @@
     "line": MATPLOTLIBLinePlot,
     "vline": MATPLOTLIBVLinePlot,
     "scatter": MATPLOTLIBScatterPlot,
+    "sequence": MATPLOTLIBSequencePlot,
     "chromatogram": MATPLOTLIBChromatogramPlot,
     "mobilogram": MATPLOTLIBMobilogramPlot,
     "spectrum": MATPLOTLIBSpectrumPlot,

pyopenms_viz/_matplotlib/core.py

@@ -16,6 +16,7 @@
     LinePlot,
     VLinePlot,
     ScatterPlot,
+    SequencePlot,
     BaseMSPlot,
     ChromatogramPlot,
     MobilogramPlot,
@@ -444,6 +445,73 @@ def plot(
             return ax, (legend_lines, legend_labels)
 
 
+class MATPLOTLIBSequencePlot(MATPLOTLIBPlot, SequencePlot):
+
+    def plot(self):
+        sequence = self.data[self.seq_col].iloc[0]
+        n_residues = len(sequence)
+
+        # Remap `x` position to be the left edge of the peptide.
+        self.x_pos = self.x_pos - n_residues * self.spacing / 2 + self.spacing / 2
+
+        # Plot the amino acids in the peptide.
+        for i, aa in enumerate(sequence):
+            self.fig.text(
+                *(self.x_pos + i * self.spacing, self.y_pos, aa),
+                fontsize=self.seq_fontsize,
+                ha="center",
+                transform=self.fig.transAxes,
+                va="center",
+            )
+        # Indicate matched fragments.
+        for annot, color in zip(
+            self.data[self.ion_annotation], self.data[self.color_annotation]
+        ):
+            ion_type = annot[0]
+            ion_i = int(i) if (i := annot[1:].rstrip("+")) else 1
+            x_i = self.x_pos + self.spacing / 2 + (ion_i - 1) * self.spacing
+
+            # Length of the fragment line.
+            if ion_type in "ax":
+                y_i = 2 * self.frag_len
+            elif ion_type in "by":
+                y_i = self.frag_len
+            elif ion_type in "cz":
+                y_i = 3 * self.frag_len
+            else:
+                # Ignore unknown ion types.
+                continue
+
+            # N-terminal fragmentation.
+            if ion_type in "abc":
+                xs = [x_i, x_i, x_i - self.spacing / 2]
+                ys = [self.y_pos, self.y_pos + y_i, self.y_pos + y_i]
+                nterm = True
+            # C-terminal fragmentation.
+            elif ion_type in "xyz":
+                xs = [x_i + self.spacing / 2, x_i, x_i]
+                ys = [self.y_pos - y_i, self.y_pos - y_i, self.y_pos]
+                nterm = False
+            else:
+                # Ignore unknown ion types.
+                continue
+
+            self.fig.plot(
+                xs, ys, clip_on=False, color=color, transform=self.fig.transAxes
+            )
+
+            self.fig.text(
+                x_i,
+                self.y_pos + (1.05 if nterm else -1.05) * y_i,
+                annot,
+                color=color,
+                fontsize=self.annotation_font_size,
+                ha="right" if nterm else "left",
+                transform=self.fig.transAxes,
+                va="top" if not nterm else "bottom",
+            )
+
+
 class MATPLOTLIB_MSPlot(BaseMSPlot, MATPLOTLIBPlot, ABC):
 
     def get_line_renderer(self, data, x, y, **kwargs) -> None:

pyopenms_viz/_plotly/__init__.py

@@ -7,6 +7,7 @@
     PLOTLYLinePlot,
     PLOTLYVLinePlot,
     PLOTLYScatterPlot,
+    PLOTLYSequencePlot,
     PLOTLYChromatogramPlot,
     PLOTLYMobilogramPlot,
     PLOTLYSpectrumPlot,
@@ -23,6 +24,7 @@
     "line": PLOTLYLinePlot,
     "vline": PLOTLYVLinePlot,
     "scatter": PLOTLYScatterPlot,
+    "sequence": PLOTLYSequencePlot,
     "chromatogram": PLOTLYChromatogramPlot,
     "mobilogram": PLOTLYMobilogramPlot,
     "spectrum": PLOTLYSpectrumPlot,

pyopenms_viz/_plotly/core.py

@@ -17,6 +17,7 @@
     LinePlot,
     VLinePlot,
     ScatterPlot,
+    SequencePlot,
     BaseMSPlot,
     ChromatogramPlot,
     MobilogramPlot,
@@ -534,6 +535,11 @@ def plot(
         return fig, None
 
 
+class PLOTLYSequencePlot(PLOTLYPlot, SequencePlot):
+    def plot(self):
+        raise NotImplementedError("Sequence plot is not implemented for Plotly")
+
+
 class PLOTLY_MSPlot(BaseMSPlot, PLOTLYPlot, ABC):
 
     def get_line_renderer(self, data, x, y, **kwargs) -> None:

sequenceplot.py

@@ -0,0 +1,83 @@
+def plot(ax, data, x=0.5, y=0.5, spacing=0.06, fontsize="xx-large", fontsize_frag="medium", frag_len=0.06):
+    """
+    Plot peptide sequence with matched fragments indicated.
+
+    Parameters
+    ----------
+    ax : matplotlib.axes.Axes
+        The axes to plot on.
+    data : pandas.DataFrame
+        The spectrum dataframe.
+    x : float, optional
+        The center horizontal position of the peptide sequence.
+    y : float, optional
+        The center vertical position of the peptide sequence.
+    spacing : float, optional
+        The horizontal spacing between amino acids.
+    fontsize : str, optional
+        The font size of the amino acids.
+    fontsize_frag : str, optional
+        The font size of the fragment annotations.
+    frag_len : float, optional
+        The length of the fragment lines.
+    """
+    sequence = data["sequence"].iloc[0]
+    n_residues = len(sequence)
+
+    # Remap `x` position to be the left edge of the peptide.
+    x = x - n_residues * spacing / 2 + spacing / 2
+
+    # Plot the amino acids in the peptide.
+    for i, aa in enumerate(sequence):
+        ax.text(
+            *(x + i * spacing, y, aa),
+            fontsize=fontsize,
+            ha="center",
+            transform=ax.transAxes,
+            va="center",
+        )
+    # Indicate matched fragments.
+    for annot, color in zip(data["ion_annotation"], data["color_annotation"]):
+        ion_type = annot[0]
+        ion_i = int(i) if (i := annot[1:].rstrip("+")) else 1
+        x_i = x + spacing / 2 + (ion_i - 1) * spacing
+
+        # Length of the fragment line.
+        if ion_type in "ax":
+            y_i = 2 * frag_len
+        elif ion_type in "by":
+            y_i = frag_len
+        elif ion_type in "cz":
+            y_i = 3 * frag_len
+        else:
+            # Ignore unknown ion types.
+            continue
+
+        # N-terminal fragmentation.
+        if ion_type in "abc":
+            xs = [x_i, x_i, x_i - spacing / 2]
+            ys = [y, y + y_i, y + y_i]
+            nterm = True
+        # C-terminal fragmentation.
+        elif ion_type in "xyz":
+            xs = [x_i + spacing / 2, x_i, x_i]
+            ys = [y - y_i, y - y_i, y]
+            nterm = False
+        else:
+            # Ignore unknown ion types.
+            continue
+
+        ax.plot(
+            xs, ys, clip_on=False, color=color, transform=ax.transAxes
+        )
+
+        ax.text(
+            x_i,
+            y + (1.05 if nterm else -1.05) * y_i,
+            annot,
+            color=color,
+            fontsize=fontsize_frag,
+            ha="right" if nterm else "left",
+            transform=ax.transAxes,
+            va="top" if not nterm else "bottom",
+        )