Merge pull request #497 from theosanderson/nexus_props

Add support for nexus props

taxonium_component/src/hooks/useSearch.jsx

@@ -37,7 +37,6 @@ const useSearch = ({
   const searchesEnabled = query.enabled
     ? JSON.parse(query.enabled)
     : JSON.parse(default_query.enabled);
-  console.log("searchesEnabled", searchesEnabled);
 
   const setEnabled = (key, enabled) => {
     console.log("setEnabled", key, enabled);

taxonium_component/src/stories/Taxonium.stories.jsx

@@ -113,3 +113,17 @@ export const LocalDataWithMetadataNew = {
     layout: "padded",
   },
 };
+
+export const NexusTree = {
+  args: {
+    sourceData: {
+      status: "url_supplied",
+      filename: "https://cov2tree.nyc3.cdn.digitaloceanspaces.com/nexus.tree",
+      filetype: "nexus",
+    },
+  },
+  parameters: {
+    // More on how to position stories at: https://storybook.js.org/docs/react/configure/story-layout
+    layout: "padded",
+  },
+};

taxonium_component/src/utils/nexusToNewick.js

@@ -1,4 +1,3 @@
-// get nexusString from tree.nexus
 function nexusToNewick(nexusString) {
   // get Translate section if present
   const translateBlock = nexusString.match(/Translate(.*?);/gims);
@@ -27,20 +26,37 @@ function nexusToNewick(nexusString) {
   // get the Newick string from the tree block
   const newickString = treeBlock[0].match(/\((.*?)\).+;/gims)[0];
 
-  //remove comments, which are indicated by [...]
+  let nodeProperties = {};
+
+  // extract properties, which are indicated by [&key=value] or [&key={value1,value2,...}]
+  newickString.replace(
+    /\[&?(.*?)\]/gims,
+    (match, contents, offset, inputString) => {
+      let nodeId = inputString.slice(0, offset).match(/[^,\(\):]+$/g)[0];
+      // use a regular expression to split on commas not inside curly brackets
+      let properties = contents.split(/,(?![^{]*})/g);
+      let propertyDict = {};
+      for (let prop of properties) {
+        let [key, value] = prop.split("=");
+        propertyDict["meta_" + key] = value;
+      }
+      nodeProperties[nodeId] = propertyDict;
+    }
+  );
+
+  // remove comments, which are indicated by [...]
 
   const newick = newickString.replace(/\[(.*?)\]/gims, "");
 
   // translate the taxon labels in the Newick string
   const translatedNewickString = newick.replace(
     /([^:\,\(\)]+)/gims,
     (match) => {
-      //console.log(translations[match])
       return translations[match] || match;
     }
   );
 
-  return translatedNewickString;
+  return { newick: translatedNewickString, nodeProperties };
 }
 
 export default nexusToNewick;

taxonium_component/src/utils/processNewick.js

@@ -117,12 +117,15 @@ async function cleanup(tree) {
 
 export async function processNewick(data, sendStatusMessage) {
   let the_data;
+  let extra_metadata;
 
   the_data = await fetch_or_extract(data, sendStatusMessage, "tree");
 
   console.log("data.filetype", data.filetype);
   if (data.filetype == "nexus") {
-    the_data = nexusToNewick(the_data);
+    const result = nexusToNewick(the_data);
+    the_data = result.newick;
+    extra_metadata = result.nodeProperties;
   }
 
   sendStatusMessage({
@@ -205,6 +208,7 @@ export async function processNewick(data, sendStatusMessage) {
     rootMutations: [],
     rootId: 0,
     overwrite_config: { num_tips: total_tips, from_newick: true },
+    extra_metadata,
   };
 
   return output;
@@ -276,29 +280,55 @@ export async function processMetadataFile(data, sendStatusMessage) {
 
 export async function processNewickAndMetadata(data, sendStatusMessage) {
   const treePromise = processNewick(data, sendStatusMessage);
-
+  let tree, metadata_double;
+  let metadata = new Map();
+  let headers = [];
   const metadataInput = data.metadata;
   if (!metadataInput) {
-    return await treePromise;
+    tree = await treePromise;
+  } else {
+    // Wait for both promises to resolve
+    [tree, metadata_double] = await Promise.all([
+      treePromise,
+      processMetadataFile(metadataInput, sendStatusMessage),
+    ]);
+    [metadata, headers] = metadata_double;
   }
-  // Wait for both promises to resolve
-  const [tree, metadata_double] = await Promise.all([
-    treePromise,
-    processMetadataFile(metadataInput, sendStatusMessage),
-  ]);
-  const [metadata, headers] = metadata_double;
+
   const blanks = Object.fromEntries(
     headers.slice(1).map((x) => ["meta_" + x, ""])
   );
+
+  if (tree.extra_metadata) {
+    // loop over the extra metadata dict to find all the (sub)keys
+    const all_extra_keys = new Set();
+    Object.values(tree.extra_metadata).forEach((node_extra) => {
+      Object.keys(node_extra).forEach((key) => {
+        all_extra_keys.add(key);
+      });
+    });
+    // add any misssing keys to blanks
+    all_extra_keys.forEach((key) => {
+      if (!blanks[key]) {
+        blanks[key] = "";
+      }
+    });
+  }
+
   sendStatusMessage({
     message: "Assigning metadata to nodes",
   });
   tree.nodes.forEach((node) => {
     const this_metadata = metadata.get(node.name);
+    Object.assign(node, blanks);
     if (this_metadata) {
       Object.assign(node, this_metadata);
-    } else {
-      Object.assign(node, blanks);
+    }
+    if (tree.extra_metadata) {
+      const node_extra = tree.extra_metadata[node.name];
+      if (node_extra) {
+        Object.assign(node, node_extra);
+      }
     }
   });