Merge pull request #134 from savitakartik/edge_span_and_area_plots

add edge area histograms
tskit-dev · Mar 28, 2024 · 948e93b · 948e93b
2 parents 26d3ee9 + 844d8ae
commit 948e93b
Show file tree

Hide file tree

Showing 4 changed files with 67 additions and 7 deletions.
diff --git a/tsqc/pages/edges.py b/tsqc/pages/edges.py
@@ -7,6 +7,7 @@
 from .. import config
 from ..plot_helpers import filter_points
 from ..plot_helpers import hover_points
+from ..plot_helpers import make_hist
 
 
 def make_edges_panel(log_y, node_type, tsm):
@@ -54,7 +55,37 @@ def make_edges_panel(log_y, node_type, tsm):
             ylabel=y_label,
         )
     )
-    return pn.Column(main)
+
+    edges_df["time_span"] = edges_df["parent_time"] - edges_df["child_time"]
+    gspan_hist = make_hist(
+        edges_df["span"],
+        "Genomic span",
+        "auto",
+        log_y=log_y,
+        xlabel="genomic span",
+        ylabel="number of edges",
+    )
+    tspan_hist = make_hist(
+        edges_df["time_span"],
+        "Time span",
+        "auto",
+        log_y=log_y,
+        xlabel="time span",
+        ylabel="number of edges",
+    )
+    area_hist = make_hist(
+        edges_df["span"] * edges_df["time_span"],
+        "Edge area",
+        "auto",
+        log_y=log_y,
+        xlabel="time span * genomic span",
+        ylabel="number of edges",
+    )
+    area_hist.opts(hv.opts.Histogram(xrotation=90))
+    gspan_hist.opts(hv.opts.Histogram(xrotation=90))
+    tspan_hist.opts(hv.opts.Histogram(xrotation=90))
+
+    return pn.Column(main, pn.Row(gspan_hist, tspan_hist, area_hist))
 
 
 def page(tsm):

diff --git a/tsqc/pages/nodes.py b/tsqc/pages/nodes.py
@@ -25,6 +25,8 @@ def make_node_hist_panel(tsm, log_y):
             bins,
             log_y=log_y,
             plot_width=config.PLOT_WIDTH,
+            xlabel="ancestor span",
+            ylabel="number of nodes",
         )
 
         return pn.Column(pn.Row(nodes_hist))

diff --git a/tsqc/pages/trees.py b/tsqc/pages/trees.py
@@ -27,6 +27,8 @@ def make_tree_hist_panel(tsm, log_y):
             bins,
             log_y=log_y,
             plot_width=config.PLOT_WIDTH,
+            xlabel="number of sites",
+            ylabel="number of trees",
         )
 
         spans_hist = make_hist(
@@ -35,6 +37,8 @@ def make_tree_hist_panel(tsm, log_y):
             bins,
             log_y=log_y,
             plot_width=config.PLOT_WIDTH,
+            xlabel="genomic span",
+            ylabel="number of trees",
         )
 
         muts_hist = make_hist(
@@ -43,6 +47,8 @@ def make_tree_hist_panel(tsm, log_y):
             bins,
             log_y=log_y,
             plot_width=config.PLOT_WIDTH,
+            xlabel="number of mutations",
+            ylabel="number of trees",
         )
 
         tbl_hist = make_hist(
@@ -51,6 +57,8 @@ def make_tree_hist_panel(tsm, log_y):
             bins,
             log_y=log_y,
             plot_width=config.PLOT_WIDTH,
+            xlabel="total branch length",
+            ylabel="number of trees",
         )
 
         mean_arity_hist = make_hist(
@@ -59,6 +67,8 @@ def make_tree_hist_panel(tsm, log_y):
             bins,
             log_y=log_y,
             plot_width=config.PLOT_WIDTH,
+            xlabel="mean arity",
+            ylabel="number of trees",
         )
 
         max_arity_hist = make_hist(
@@ -67,6 +77,8 @@ def make_tree_hist_panel(tsm, log_y):
             bins,
             log_y=log_y,
             plot_width=config.PLOT_WIDTH,
+            xlabel="max arity",
+            ylabel="number of trees",
         )
         return pn.Column(
             pn.Row(

diff --git a/tsqc/plot_helpers.py b/tsqc/plot_helpers.py
@@ -34,21 +34,36 @@ def compute_hist(x_range, y_range):
     return compute_hist
 
 
-def make_hist(data, title, bins_range, log_y=True, plot_width=800):
+def make_hist(
+    data,
+    title,
+    bins_range,
+    xlabel,
+    ylabel="count",
+    log_y=True,
+    plot_width=400,
+    plot_height=400,
+):
     """
     Make histogram from given count data
     """
     count, bins = np.histogram(data, bins=bins_range)
-    ylabel = "log(Count)" if log_y else "Count"
     np.seterr(divide="ignore")
     if log_y:
         count = np.log10(count)
         count[count == -np.inf] = 0
+        ylabel = f"log({ylabel})"
+
     histogram = hv.Histogram((count, bins)).opts(
-        title=title, ylabel=ylabel, tools=["hover"]
-    )
-    histogram = histogram.opts(
-        shared_axes=False, width=round(plot_width / 2), toolbar=None, default_tools=[]
+        tools=["hover"],
+        shared_axes=False,
+        width=plot_width,
+        height=plot_height,
+        toolbar=None,
+        default_tools=[],
+        title=title,
+        xlabel=xlabel,
+        ylabel=ylabel,
     )
     return histogram