+ heatmaps hauto, hmin and hmax options

Plots/paf_to_heatmap.py

@@ -1,7 +1,7 @@
 #!/usr/bin/env python3
 ## Pombert lab, 2024
-version = '0.1b'
-updated = '2024-05-01'
+version = '0.2'
+updated = '2024-05-03'
 name = 'paf_to_heatmap.py'
 
 import sys
@@ -35,11 +35,14 @@
 -o (--outdir)   Output directory [Default: ./]
 -h (--height)   Figure height in inches [Default: 10]
 -w (--width)    Figure width in inches [Default: 10]
+-x (--matrix)   Matrix output file [Default: matrix.tsv]
 -c (--palette)  Seaborn color palette [Default: winter_r]
                 # See https://www.practicalpythonfordatascience.com/ap_seaborn_palette
                 # for a list of color palettes
 --fontsize      Font size [Default: 8]
--x (--matrix)   Matrix output file [Default: matrix.tsv]
+--vmax          Set maximum color bar value [Default: 100]
+--vmin          Set minimum color bar value [Default: 0]
+--vauto         Set color bar values automatically instead
 """
 
 # Print custom message if argv is empty
@@ -60,16 +63,22 @@
 cmd.add_argument("-c", "--palette", default='winter_r')
 cmd.add_argument("-x", "--matrix", default='matrix.tsv')
 cmd.add_argument("--fontsize", default=8)
+cmd.add_argument("--vmin", default=0)
+cmd.add_argument("--vmax", default=100)
+cmd.add_argument("--vauto", action='store_true')
 args = cmd.parse_args()
 
 paf_files = args.paf
 fasta_files = args.fasta
 outdir = args.outdir
 height = args.height
 width = args.width
+matrix_file = args.matrix
 color_palette = args.palette
 fontsize = int(args.fontsize)
-matrix_file = args.matrix
+vmin = int(args.vmin)
+vmax = int(args.vmax)
+vauto = args.vauto
 
 ################################################################################
 ## Working on output directory
@@ -208,14 +217,25 @@
     heatmap_svg = svgdir + '/' + 'colinear_bases.mmap.heatmap.svg'
 
     ## Clustered heatmaps
-    cm = sns.clustermap(
-        data[0:],
-        cmap=color_palette,
-        annot=True,
-        fmt='.1f',
-        vmin=0,
-        vmax=100
-    )
+    cm = None
+
+    if vauto:
+        cm = sns.clustermap(
+            data[0:],
+            cmap=color_palette,
+            annot=True,
+            fmt='.1f'
+        )
+
+    else:
+        cm = sns.clustermap(
+            data[0:],
+            cmap=color_palette,
+            annot=True,
+            fmt='.1f',
+            vmin=vmin,
+            vmax=vmax
+        )
 
     cm.fig.suptitle(f"% of total bases in pairwise alignments", x=0.5, y=0.95)
     print(f"1 / 4 - Plotting {clustered_png}")
@@ -227,14 +247,25 @@
     plt.close('all')
 
     ## Normal heatmaps
-    hm = sns.heatmap(
-        data[0:],
-        cmap=color_palette,
-        annot=True,
-        fmt='.1f',
-        vmin=0,
-        vmax=100
-    )
+    hm = None
+
+    if vauto:
+        hm = sns.heatmap(
+            data[0:],
+            cmap=color_palette,
+            annot=True,
+            fmt='.1f'
+        )
+
+    else:
+        hm = sns.heatmap(
+            data[0:],
+            cmap=color_palette,
+            annot=True,
+            fmt='.1f',
+            vmin=vmin,
+            vmax=vmax
+        )
 
     hm.figure.suptitle(f"% of colinear bases in pairwise alignments", x=0.5, y=0.95)
     print(f"3 / 4 - Plotting {heatmap_png}")

Plots/protein_cluster_hm.py

@@ -2,8 +2,8 @@
 ## Pombert lab, 2024
 
 name = 'protein_cluster_hm.py'
-version = '0.3c'
-updated = '2024-05-01'
+version = '0.3d'
+updated = '2024-05-03'
 
 import sys
 import os
@@ -37,6 +37,9 @@
 -h (--height)   Figure height in inches [Default: 10]
 -w (--width)    Figure width in inches [Default: 10]
 --fontsize      Font size [Default: 8]
+--vmax          Set maximum color bar value [Default: 100]
+--vmin          Set minimum color bar value [Default: 0]
+--vauto         Set color bar values automatically instead
 --threads       Number of threads to use [Default: 16]
 """
 
@@ -56,6 +59,9 @@
 cmd.add_argument("-w", "--width", default=10)
 cmd.add_argument("-p", "--palette", default='winter_r')
 cmd.add_argument("--fontsize", default=8)
+cmd.add_argument("--vmin", default=0)
+cmd.add_argument("--vmax", default=100)
+cmd.add_argument("--vauto", action='store_true')
 cmd.add_argument("--threads", default=16)
 args = cmd.parse_args()
 
@@ -65,6 +71,9 @@
 width = args.width
 color_palette = args.palette
 fontsize = int(args.fontsize)
+vmin = int(args.vmin)
+vmax = int(args.vmax)
+vauto = args.vauto
 threads = int(args.threads)
 
 ################################################################################
@@ -111,14 +120,25 @@ def heatmap(tsv_file):
         heatmap_svg = svgdir + '/' + 'proteins_in_clusters.gap_' + gap + '.heatmap.svg'
 
         ## Clustered heatmaps
-        cm = sns.clustermap(
-            data[0:],
-            cmap=color_palette,
-            annot=True,
-            fmt='.1f',
-            vmin=0,
-            vmax=100,
-        )
+        cm = None
+
+        if vauto:
+            cm = sns.clustermap(
+                data[0:],
+                cmap=color_palette,
+                annot=True,
+                fmt='.1f'
+            )
+
+        else:
+            cm = sns.clustermap(
+                data[0:],
+                cmap=color_palette,
+                annot=True,
+                fmt='.1f',
+                vmin=vmin,
+                vmax=vmax
+            )
 
         cm.fig.suptitle(f"% of protein-coding genes in clusters (gap = {gap})", x=0.5, y=0.95)
         with counter.get_lock():
@@ -133,14 +153,25 @@ def heatmap(tsv_file):
         plt.close('all')
 
         ## Normal heatmaps
-        hm = sns.heatmap(
-            data[0:],
-            cmap=color_palette,
-            annot=True,
-            fmt='.1f',
-            vmin=0,
-            vmax=100
-        )
+        hm = None
+
+        if vauto:
+            hm = sns.heatmap(
+                data[0:],
+                cmap=color_palette,
+                annot=True,
+                fmt='.1f'
+            )
+
+        else:
+            hm = sns.heatmap(
+                data[0:],
+                cmap=color_palette,
+                annot=True,
+                fmt='.1f',
+                vmin=vmin,
+                vmax=vmax
+            )
 
         hm.figure.suptitle(f"% of proteins found in clusters (gap = {gap})", x=0.5, y=0.95)
         with counter.get_lock():

README.md

@@ -286,6 +286,9 @@ Options for run_SYNY.pl are:
 -hw (--hwidth)          Heatmap figure width in inches [Default: 10]
 --hfsize                Heatmap font size [Default: 8]
 --hmpalette             Heatmap color palette [Default: winter_r]
+--hmax                  Set maximum color bar value [Default: 100]
+--hmin                  Set minimum color bar value [Default: 0]
+--hauto                 Set color bar values automatically instead
 --no_heatmap            Turn off heatmaps
 ```
 The output directory will be structured as follows: 
@@ -535,7 +538,7 @@ In the above example, small heatmaps with 25 datapoints (5 * 5 genomes) will be
   <img src="https://github.com/PombertLab/SYNY/blob/main/Images/proteins_in_clusters.gap_0.heatmap.png">
 </p>
 
-Heatmap dimensions (default: 10 x 10) can be modified with the `--hheight` and `--hwidth` command line switches. The color palette (default: winter_r) can be modified with the `--hmpalette` command line switch (see this [URL](https://github.com/PombertLab/SYNY/blob/main/Images/python_color_palettes.png) for a list of color palettes). Color palettes available on the operating system can be listed and/or plotted with [check_mp_colors.py](https://github.com/PombertLab/SYNY/blob/main/Utils/check_mp_colors.py).
+Heatmap dimensions (default: 10 x 10) can be modified with the `--hheight` and `--hwidth` command line switches. Color bar min/max values (defaults: 0/100) can be changed with the `--hmin` and `--hmax` command line switches or calculated automatically with `--hauto`. The heatmap color palette (default: winter_r) can be changed with the `--hmpalette` command line switch (see this [URL](https://github.com/PombertLab/SYNY/blob/main/Images/python_color_palettes.png) for a list of color palettes). Color palettes available on the operating system can be listed and/or plotted with [check_mp_colors.py](https://github.com/PombertLab/SYNY/blob/main/Utils/check_mp_colors.py).
 
 #### Circos plots