tweaks

MrTomRod · Apr 27, 2022 · 45e0268 · 45e0268
1 parent 083130b
commit 45e0268
Show file tree

Hide file tree

Showing 6 changed files with 70 additions and 43 deletions.
diff --git a/scoary/create_final_overview.py b/scoary/create_final_overview.py
@@ -10,12 +10,13 @@
 from matplotlib.figure import Figure
 from matplotlib.collections import PatchCollection, QuadMesh
 from matplotlib.patches import Rectangle
-from matplotlib.colors import LogNorm
+from matplotlib.colors import LogNorm, ListedColormap, Colormap, LinearSegmentedColormap
 
 from .utils import MockNamespace, ROOT_DIR
 
 
 def plot_dendrogram(linkage_matrix: np.ndarray, labels: [str], ax: Axes) -> {}:
+    # no recursion problem here. Works with up to 20'000 isolates, becomes more of a memory problem.
     dendrogram_params = hierarchy.dendrogram(
         linkage_matrix,
         orientation='left',
@@ -53,12 +54,12 @@ def add_clickable_patches(patch_names, fig: Figure, ax: Axes):
     fig.add_artist(pc)
 
 
-def plot_qvals(qvals: pd.DataFrame, fig: Figure, ax: Axes, cmap: str = None) -> [QuadMesh]:
+def plot_qvals(qvals: pd.DataFrame, fig: Figure, ax: Axes, cmaps: {str: str | Colormap}) -> [QuadMesh]:
     # determine y intervals: [0, 10, 20, ...]
     y = np.arange(start=0, stop=len(qvals.index) * 10 + 1, step=10)
 
     pcms = []
-    for i, col in enumerate(qvals.columns):
+    for i, (col, cmap) in enumerate(cmaps.items()):
         # determine x intervals: [0, 1] / [1, 2] / ...
         x = np.array([i, i + 1])
 
@@ -161,10 +162,15 @@ def create_final_overview(overview_ds: pd.DataFrame, ns: MockNamespace, isolate_
     overview_ds = overview_ds.reindex(dendrogram_params['ivl'])
 
     # plot qvals
-    cols = ['min_qval', 'min_pval_empirical']
-    cols = [c for c in cols if c in overview_ds.columns]
-
-    pcms = plot_qvals(overview_ds[cols], fig=fig, ax=ax_colorbar, cmap='afmhot')
+    cmaps = {
+        'min_qval': 'Spectral',
+        'min_pval_empirical': LinearSegmentedColormap.from_list(
+            name='pval_emp_cbar',
+            colors=['#590d22', '#800f2f', '#a4133c', '#c9184a', '#ff4d6d',
+                    '#ff758f', '#ff8fa3', '#ffb3c1', '#ffccd5', '#fff0f3'])
+    }
+    cols = [col for col in cmaps.keys() if col in overview_ds.columns]
+    pcms = plot_qvals(overview_ds[cols], fig=fig, ax=ax_colorbar, cmaps=cmaps)
 
     # save plot
     plt.savefig(f'{ns.outdir}/overview_plot.svg', format='svg')

diff --git a/scoary/scoary.py b/scoary/scoary.py
@@ -47,7 +47,7 @@ def scoary(
     )
 
     # dynamically set recursion limit, should work for ~ 13'000 isolates
-    _recursion_limit = 100 + len(traits_df.index) ** 2
+    _recursion_limit = max(1000, 100 + len(traits_df.index) ** 2)
     logger.warning(f'Setting recursion limit to {_recursion_limit}')
     sys.setrecursionlimit(_recursion_limit)
 

diff --git a/scoary/templates/config.json b/scoary/templates/config.json
@@ -1,6 +1,4 @@
 {
-  "default-hidden-cols": ["sensitivity", "specificity", "odds_ratio", "contrasting", "supporting", "opposing", "best", "worst", "pval"],
-  "sanitize-genes": true,
   "colors": {
     "g+t+": "#1f78b4",
     "g+t-": "#a6cee3",
@@ -9,20 +7,25 @@
     "g-t-": "#33a02c",
     "g-t?": "#edffdf"
   },
+  "table-config": {
+    "sanitize-genes": true,
+    "default-hidden-cols": ["sensitivity", "specificity", "odds_ratio", "contrasting", "supporting", "opposing", "best", "worst", "pval"],
+    "float-cols": ["pval", "qval", "pval_empirical", "best", "worst"]
+  },
   "tree-config": {
     "type": "Hierarchical",
     "height": 400,
     "leaf-nodes": {
       "g+": {"shape": "hexagon", "fillColour": "#000000"},
       "g-": {"shape": "diamond", "fillColour": "#e3e3e3"}
     },
-    "color-scale": ["yellow", "red"],
-    "metadata-bars": {
-      "g+": {"colour": "#000000", "label": "present"},
-      "g-": {"colour": "#e3e3e3", "label": "absent"},
-      "t+": {"colour": "red", "label": "present"},
-      "t-": {"colour": "yellow", "label": "absent"},
-      "t?": {"colour": "#ffffff", "label": "unclear"}
+      "color-scale": ["yellow", "red"],
+      "metadata-bars": {
+        "g+": {"colour": "#000000", "label": "present"},
+        "g-": {"colour": "#e3e3e3", "label": "absent"},
+        "t+": {"colour": "red", "label": "present"},
+        "t-": {"colour": "yellow", "label": "absent"},
+        "t?": {"colour": "#ffffff", "label": "unclear"}
     }
   },
   "link-config": {

diff --git a/scoary/templates/trait.js b/scoary/templates/trait.js
@@ -103,7 +103,7 @@ documentReadyPromise.then(() => document.querySelector('#trait-name').textConten
 const configPromise = fetch(`config.json`)
     .then(response => response.json())
     .then(config => {
-        if (config['sanitize-genes']) {
+        if (config['table-config']['sanitize-genes']) {
             config.geneListToArray = (d, i) => d[i].split(',').map((gene) => gene.split('|').slice(-1)[0])
         } else {
             config.geneListToArray = (d, i) => d[i].split(',')
@@ -119,6 +119,13 @@ const metaPromise = fetch(`traits/${trait}/meta.json`)
     .then(response => response.json())
 
 
+/**
+ * Promise for tree.nwk. Returns the dictionary object.
+ */
+const newickPromise = fetch('tree.nwk')
+    .then(response => response.text())
+
+
 /**
  * Show metadata.
  */
@@ -193,13 +200,18 @@ const tablePromise = configPromise.then(config => {
         header: true, download: true, skipEmptyLines: true, delimiter: '\t', newline: '\n',
     }).then(tableData => {
         // Preprocess data
+        const tableConfig = config['table-config']
+        const floatCols = tableConfig['float-cols']
+        const renderFloat = (data, type, row) => parseFloat(data).toPrecision(3)
         tableData.columns = tableData.meta.fields.map(col => {
-            return {'data': col, 'label': col, 'title': col}
+            const column = {'data': col, 'label': col, 'title': col}
+            if (floatCols.includes(col)) column['render'] = renderFloat
+            return column
         })
         tableData.index = tableData['data'].map(col => col['Gene'])
         tableData.orderCol = tableData.meta.fields.indexOf(
             tableData.meta.fields.includes('pval_empirical') ? 'pval_empirical' : 'qval')
-        tableData.hiddenCols = config['default-hidden-cols'].map(col => tableData.meta.fields.indexOf(col))
+        tableData.hiddenCols = tableConfig['default-hidden-cols'].map(col => tableData.meta.fields.indexOf(col))
 
         return tableData
     })
@@ -264,7 +276,7 @@ const isolateInfoPromise = Papa.execPromise('isolate_info.tsv', {
 
     return isolateInfo
 }).catch(() => {
-    console.log('no isolate info')
+    console.info('no isolate info')
     return 'no isolate info'
 })
 
@@ -378,18 +390,17 @@ const coverageMatrixPromise = Promise.all(
  * Hide all shown popovers if a click on a non-popover is registered.
  */
 let popovers = []
-$(document).on("click", (event) => {
-    // ignore clicks on .popover or .has-popover
-    if (Boolean(event.target.closest('.popover, .has-popover'))) {
-        return
-    }
-
-    // hide all popovers
+const hidePopovers = () => {
     while (popovers.length > 0) {
         const popover = popovers.pop()
         bootstrap.Popover.getInstance(popover).hide()
     }
-
+}
+$(document).on("click", (event) => {
+    // ignore clicks on .popover or .has-popover
+    if (!Boolean(event.target.closest('.popover, .has-popover'))) {
+        hidePopovers()
+    }
 })
 
 
@@ -439,6 +450,7 @@ const coverageMatrixTablePromise = Promise.all(
         return `${gene} x ${isolate}`
     }
 
+    if (!cmValues.isNumeric) return cmValues
 
     /**
      * Make genes clickable depending on config.json.
@@ -609,7 +621,7 @@ Promise.all([tablePromise, coverageMatrixTablePromise, valuesPromise, configProm
 Promise.all([tablePromise, metaPromise, valuesPromise, coverageMatrixPromise, configPromise])
     .then(([tableData, metaData, valuesData, cmValues, config]) => {
         if (!valuesData.isNumeric) {
-            console.log('no numeric data')
+            console.info('no numeric data')
             return
         }
 
@@ -706,6 +718,8 @@ Promise.all([coverageMatrixPromise, valuesPromise, configPromise])
         }
 
         const genesPopover = (event, currentGene) => {
+            if (popovers.includes(event.target)) throw 'popover already exists'
+
             event.target.classList.add('has-popover')
 
             new bootstrap.Popover(event.target, {
@@ -721,27 +735,22 @@ Promise.all([coverageMatrixPromise, valuesPromise, configPromise])
 
         // append to click event
         geneClickFunctions.push(genesPopover)
-
-
     })
 
 
-// load newick file (tree)
-const newickPromise = fetch('tree.nwk')
-    .then(response => response.text())
-
-
-const getSize = (element) => {
+const getInnerWidth = (element) => {
     const computedStyle = getComputedStyle(element)
     return element.clientWidth - parseFloat(computedStyle.paddingLeft) - parseFloat(computedStyle.paddingRight)
 }
+
+
 // draw tree
 const treePromise = Promise.all([documentReadyPromise, newickPromise, valuesPromise, configPromise])
     .then(([_, newickTree, valuesData, config]) => {
         // https://www.phylocanvas.gl/examples/metadata-blocks.html
         // https://www.phylocanvas.gl/docs/methods.html
         const treeContainer = document.querySelector("#tree")
-        let currentWidth = getSize(treeContainer.parentElement)
+        let currentWidth = getInnerWidth(treeContainer.parentElement)
         const tree = new phylocanvas.PhylocanvasGL(
             treeContainer, {
                 type: phylocanvas.TreeTypes[config['tree-config']['type']],
@@ -759,7 +768,7 @@ const treePromise = Promise.all([documentReadyPromise, newickPromise, valuesProm
         )
 
         const redrawWrapper = () => {
-            const newWidth = getSize(treeContainer.parentElement)
+            const newWidth = getInnerWidth(treeContainer.parentElement)
             if (newWidth !== currentWidth) {
                 tree.resize(newWidth, config['tree-config']['height'])
                 currentWidth = newWidth
@@ -862,7 +871,7 @@ const calcOrthogeneData = (gene, coverageMatrix, valuesData) => {
         const isolate = isolateData['Isolate']
         const genePositive = isolateData[sanitizedGene].toString() !== '0'
         const traitPositive = valuesData.isolateToTrait[isolate]
-        const traitValue = valuesData.isolateToValue[isolate]
+        const traitValue = valuesData.isNumeric ? valuesData.isolateToValue[isolate] : null
 
         if (genePositive) {
             if (traitPositive === 'true') {

diff --git a/scoary/utils.py b/scoary/utils.py
@@ -180,7 +180,8 @@ def load_info_file(
 
     if expected_overlap_set is not None:
         overlap_size = len(set.intersection(set(info_df.index), expected_overlap_set))
-        assert overlap_size > 0, f'The {merge_col}s in {info_file} do not match any {merge_col}s in {reference_file}'
+        if overlap_size == 0:
+            logger.warning(f'The {merge_col}s in {info_file} do not match any {merge_col}s in {reference_file}')
         logger.info(f'Loaded descriptions for {overlap_size} {merge_col}s')
 
     logger.info(f'Loaded {merge_col} descriptions. columns={info_df.columns.tolist()}')

diff --git a/tests/init_tests.py b/tests/init_tests.py
@@ -50,7 +50,15 @@
         'traits-lc-meta': 'LC-meta.tsv',
         'traits-gc-vol': 'GC-VOL.tsv',
         'traits-gc-vol-meta': 'none-meta.tsv',
-    }
+    },
+    'full_ds': {
+        'genes': 'N0.tsv',
+        'gene-info': 'N0_best_names.tsv',
+        'isolate-info': 'isolate_info.tsv',
+        'traits': 'traits.tsv',
+        'trait-info': 'trait_info.tsv',
+    },
+
 }
 for tree_id in range(5):
     DATA['bigger_ds'][f'genes-{tree_id}'] = f'pres_abs_tree_id_{tree_id}.csv'