Xbbtree

.gitignore

@@ -2,6 +2,10 @@
 !.github
 data/*
 data_test/*
+!data_test/timetree.nwk
+!data_test/xbbtree.nwk
+!data_test/clusters.json
+!data_test/dbstats.json
 !data/*mut_annotations.json
 .idea/
 treetime/*

batch.py

@@ -149,13 +149,60 @@ def process_feed(args, callback=None):
     # filter data, align genomes, extract features, sort by lineage
     by_lineage = process_feed(args, cb.callback)
 
-    # reconstruct time-scaled tree relating lineages
-    timetree, residuals = build_timetree(by_lineage, args, cb.callback)
+    # separate XBB and other recombinant lineages
+    aliases = parse_alias(args.alias)
+    designation = {}
+    for prefix, truename in aliases.items():
+        if type(truename) is list:
+            designation.update({prefix: {
+                'type': 'XBB' if prefix == 'XBB' else 'recombinant',
+                'fullname': '/'.join(truename)
+            }})
+        else:
+            designation.update({prefix: {
+                'type': 'XBB' if truename.startswith("XBB") else 'non-recombinant',
+                'fullname': truename
+            }})
+
+    # use results to partition by_lineage database
+    non_recomb = {}
+    xbb = {}
+    other_recomb = {}
+    for lineage, ldata in by_lineage.items():
+        # Put unassigned lineages in non-recombinant category
+        if lineage == "Unassigned":
+            non_recomb.update({lineage: ldata})
+            continue
+
+        prefix = lineage.split('.')[0]
+        category = designation[prefix]['type']
+        if category == 'non-recombinant':
+            non_recomb.update({lineage: ldata})
+        elif category == 'XBB':
+            xbb.update({lineage: ldata})
+        else:
+            other_recomb.update({lineage: ldata})
+
+    if len(xbb) < 2:
+        other_recomb.update(xbb)
+        xbb = None  # no point in building a tree
+
+
+    # reconstruct time-scaled trees 
+    timetree, residuals = build_timetree(non_recomb, args, cb.callback)
     timestamp = datetime.now().isoformat().split('.')[0]
     nwk_file = os.path.join(args.outdir, 'timetree.{}.nwk'.format(timestamp))
     with open(nwk_file, 'w') as handle:
         Phylo.write(timetree, file=handle, format='newick')
 
+    xbb_file = os.path.join(args.outdir, 'xbbtree.{}.nwk'.format(timestamp))
+    with open(xbb_file, 'w') as handle:
+        if xbb is not None:
+            timetree_xbb, residuals_xbb = build_timetree(xbb, args, cb.callback)
+            residuals.update(residuals_xbb)
+            Phylo.write(timetree_xbb, file=handle, format='newick')
+        # else empty file
+
     # clustering analysis of lineages
     result, infection_prediction = make_beadplots(by_lineage, args, cb.callback, t0=cb.t0.timestamp())
     clust_file = os.path.join(args.outdir, 'clusters.{}.json'.format(timestamp))
@@ -172,9 +219,7 @@ def process_feed(args, callback=None):
     # write data stats
     dbstat_file = os.path.join(args.outdir, 'dbstats.{}.json'.format(timestamp))
 
-    alias = parse_alias(args.alias)
-
-    with open(dbstat_file, 'w') as handle:
+    with (open(dbstat_file, 'w') as handle):
         # total number of sequences
         nseqs = 0
         for records in by_lineage.values():
@@ -187,7 +232,14 @@ def process_feed(args, callback=None):
         }
         for lineage, records in by_lineage.items():
             prefix = lineage.split('.')[0]
-            lname = lineage.replace(prefix, alias[prefix]) if lineage.lower() not in ['unclassifiable', 'unassigned'] and not prefix.startswith('X') and alias[prefix] != '' else lineage
+
+            # resolve PANGO prefix aliases 
+            lname = lineage
+            if (lineage.lower() not in ['unclassifiable', 'unassigned'] 
+                    and not prefix.startswith('X') 
+                    and aliases[prefix] != ''):
+                lname = lineage.replace(prefix, aliases[prefix])
+
             samples = unpack_records(records)
             ndiffs = [len(x['diffs']) for x in samples]
             val['lineages'][lineage] = {

cypress/e2e/beadplot.spec.cy.js

@@ -3,21 +3,19 @@ import clusters from '../../data_test/clusters.json'
 describe('Beadplot components', () => {
     it('Edge Labels', ()=> {
         cy.get('.beadplot-content>svg>g>text').should(($text) => {
-            expect($text).to.have.length(3)
-            expect($text.eq(0)).to.contain('unsampled0')
-            expect($text.eq(1)).to.contain('Scotland/GCVR-16F930')
-            expect($text.eq(2)).to.contain('Scotland/GCVR-170556')
+            expect($text).to.have.length(14)
+            expect($text.eq(0)).to.contain('USA/AZ-CDC-LC0932884')
+            expect($text.eq(1)).to.contain('USA/AZ-CDC-STM-TNFPTQHEE')
+            expect($text.eq(2)).to.contain('USA/AZ-CDC-LC0951048')
         }) 
     })
     it('Samples', () => {
         cy.get('.beadplot-content>svg>g>circle').then(($circ) => {
-            expect($circ).to.have.length(4)
-            for (let i = 0; i < 3; i++) {
-                cy.get($circ.eq(i)).should('have.attr', 'cy', 45)
+            expect($circ).to.have.length(23)
+            for (let i = 0; i < 7; i++) {
+                cy.get($circ.eq(i)).should('have.attr', 'cy', 20)
             }
-            cy.get($circ.eq(3)).should('have.attr', 'cy', 70)
-        }) 
-
+        })
     })
 })
 
@@ -29,19 +27,15 @@ describe('Tables', () => {
     })
     it('Country details: gentable', () => {
         cy.get('#tabs').contains('Countries').click()
-        var tips = {'allregions': {'North America': 13}, 'country': {'Canada': 8, 'USA': 5}}
+        var tips = {'allregions': {'North America': 23}, 'country': {'USA': 23}}
 
         cy.window().then((win) => {
             win.gentable(tips)
         })
-        cy.get('table:visible>tbody>tr').should('have.length', 2)
+        cy.get('table:visible>tbody>tr').should('have.length', 1)
         cy.get('table:visible>tbody>tr').eq(0).contains('North America')
-        cy.get('table:visible>tbody>tr').eq(0).contains('Canada')
-        cy.get('table:visible>tbody>tr').eq(0).contains('8')
-
-        cy.get('table:visible>tbody>tr').eq(1).contains('North America')
-        cy.get('table:visible>tbody>tr').eq(1).contains('USA')
-        cy.get('table:visible>tbody>tr').eq(1).contains('5')
+        cy.get('table:visible>tbody>tr').eq(0).contains('USA')
+        cy.get('table:visible>tbody>tr').eq(0).contains('23')
     })
 })
 
@@ -60,7 +54,7 @@ describe('Edge slider', () => {
 
         // https://stackoverflow.com/questions/64855669/moving-slider-with-cypress
 
-        let targetValue = 12
+        let targetValue = 5.01
         let currentValue = 2
         let increment = 0.01
         steps = (targetValue - currentValue) / increment + 1
@@ -74,28 +68,23 @@ describe('Edge slider', () => {
         cy.get('#custom-handle').should('contain', targetValue)
     })
     it('All edges are visible on max slider value ', () => {
-        cy.get('[stroke="#bbd"]').should('have.length', 2)
-    })
-    it('No edges are visible on slider value of 11.99 ', () => {
-        cy.get('#left-arrow').click()
-        cy.get('#custom-handle').should('contain', 11.99)
-        cy.get('[stroke="#bbd"]').should('not.exist')
+        cy.get('[stroke="#bbd"]').should('have.length', 12)
     })
 })
 
 describe('Tooltips', () => {
     it('Horizontal Edge', () => {
-        cy.get('#Scotland-GCVR-16F930').first().trigger('mouseover')
-        cy.get('.tooltip').contains('Parent: unsampled0')
-        cy.get('.tooltip').contains('Europe: 3')
-        cy.get('.tooltip').contains('Unique collection dates: 3')
-        cy.get('.tooltip').contains('2020-03-19 / 2020-03-27')
+        cy.get('#USA-AZ-CDC-LC0983866').first().trigger('mouseover')
+        cy.get('.tooltip').contains('Parent: USA/AZ-CDC-LC0932884')
+        cy.get('.tooltip').contains('North America: 2')
+        cy.get('.tooltip').contains('Unique collection dates: 2')
+        cy.get('.tooltip').contains('2023-01-02 / 2023-01-07')
     })
     it('Bead', () => {
-        cy.get('circle:visible').first().trigger('mouseover')
+        cy.get('#EPI_ISL_1054790').first().trigger('mouseover')
         cy.get('.tooltip').contains('Parent: unsampled0')
-        cy.get('.tooltip').contains('Genomic distance: 12')
-        cy.get('.tooltip').contains('Europe: 1')
-        cy.get('.tooltip').contains('Collection date: 2020-03-19')
+        cy.get('.tooltip').contains('Genomic distance: 1.82')
+        cy.get('.tooltip').contains('North America: 1')
+        cy.get('.tooltip').contains('Collection date: 2022-11-12')
     })
 })

cypress/e2e/covizu.spec.cy.js

@@ -89,9 +89,9 @@ describe('Search Interface', () => {
         })
 
         // this test was failing because (i think) it was searching for a data-point ('HMH') that no longer exists in the current dataset
-        // I've replaced 'HMH' with 'China' 
-        it("Searching 'China' results in selection and expected point count", () => {
-            cy.get('#search-input').type('China')
+        // I've replaced 'HMH' with 'Japan'
+        it("Searching 'Japan' results in selection and expected point count", () => {
+            cy.get('#search-input').type('Japan')
             cy.get('#search-button').click({force:true})
             cy.get('.selectionH').should('exist')
             cy.get('.SelectedCluster').should('exist')
@@ -123,20 +123,28 @@ describe('Search Interface', () => {
 
 describe('Tooltips', () => {
     it('Appear on hover over cluster', () => {
-        cy.get('[id=id-0]').trigger('mouseover');
-        cy.get('.tooltip').should('be.visible').should('have.length', 1)
+        cy.window().then((win)=>{
+            win.reset_tree(true);
+            const tip_id = win.display_id.non_recombinants.last;
+            cy.get(`[id=id-${tip_id}]`).trigger('mouseover');
+            cy.get('[id=id-563]').trigger('mouseover');
+            cy.get('.tooltip').should('be.visible').should('have.length', 1)
+        });
     })
     it('Cluster tooltips contain relevant and correct information', () => {
-        cy.get('[id=id-0]').trigger('mouseover');
         cy.window().then((win)=>{
-            cy.get('.tooltip').contains(`Sampled: ${win.tips[0]['varcount']}`)
-            cy.get('.tooltip').contains(`Displayed: ${win.tips[0]['sampled_varcount']}`)
-            cy.wrap(Object.keys(win.tips[0]['allregions'])).each(($el, index) => {
-                cy.get('.tooltip').contains(`${$el}: ${Object.values(win.tips[0]['allregions'])[index]}`)
+            win.reset_tree(true);
+            const tip_id = win.display_id.non_recombinants.last;
+            const last_index = win.tips.length - 1;
+            cy.get(`[id=id-${tip_id}]`).trigger('mouseover');
+            cy.get('.tooltip').contains(`Sampled: ${win.tips[last_index]['varcount']}`)
+            cy.get('.tooltip').contains(`Displayed: ${win.tips[last_index]['sampled_varcount']}`)
+            cy.wrap(Object.keys(win.tips[last_index]['allregions'])).each(($el, index) => {
+                cy.get('.tooltip').contains(`${$el}: ${Object.values(win.tips[last_index]['allregions'])[index]}`)
             })
-            cy.get('.tooltip').contains(`Mean diffs from root: ${Math.round(100*win.tips[0]['mean_ndiffs'])/100}`) 
-            cy.get('.tooltip').contains(`Deviation from clock: ${win.tips[0]['residual'].toFixed(2)}`)
-            cy.get('.tooltip').contains(`${win.tips[0]['first_date'].toISOString().slice(0, 10)} / ${win.tips[0]['last_date'].toISOString().slice(0, 10)}`)
+            cy.get('.tooltip').contains(`Mean diffs from root: ${Math.round(100*win.tips[last_index]['mean_ndiffs'])/100}`)
+            cy.get('.tooltip').contains(`Deviation from clock: ${win.tips[last_index]['residual'].toFixed(2)}`)
+            cy.get('.tooltip').contains(`${win.tips[last_index]['first_date'].toISOString().slice(0, 10)} / ${win.tips[last_index]['last_date'].toISOString().slice(0, 10)}`)
         })
     })
     it('Appear on hover over bead', () => {
@@ -182,7 +190,7 @@ describe("Colour tree", () => {
         let region_color_map = {};
         cy.get("#select-tree-colours").select(0).should('have.value','Region')
         cy.window().then((win)=>{
-            regions = [...Array.from(new Set(Object.values(win.region_map))),"China"]
+            regions = [...Array.from(new Set(Object.values(win.region_map))),"China", "South America", "Oceania"]
             return regions;
         })
         .then(()=>{
@@ -214,14 +222,14 @@ describe("Colour tree", () => {
             // check and see if this region_color_map is correctly represented in the tree graph
             cy.window().then((win)=>{
                 let region_title;
-                const tip_id = 215;
+                win.reset_tree(true);
+                const tip_id = win.display_id.non_recombinants.last;
                 cy.get(`[id=id-${tip_id}]`).should('be.visible').trigger('mouseover');
                 cy.get(`[id=id-${tip_id}]`).invoke('css','fill').then((tip_color)=>{
-                    region_title = Object.keys(win.tips[tip_id]['allregions'])[0];
+                    region_title = Object.keys(win.tips[win.tips.length - 1]['allregions'])[0];
                     // expect(tip_color).to.equal(region_color_map[region_title])
                 })
             })
         })
     })
-
 })

cypress/e2e/utils.spec.cy.js

@@ -13,8 +13,8 @@ describe('isLineage Function', () => {
     it('Retruns true if the string is a lineage ', () => {
         cy.window().then((win) => {
             expect(win.isLineage("B.101.511")).to.eq(false)
-            expect(win.isLineage("B.4.1")).to.eq(true)
-            expect(win.isLineage("B.1.446")).to.eq(true)
+            expect(win.isLineage("BA.5.3")).to.eq(true)
+            expect(win.isLineage("CZ.2")).to.eq(true)
         })
     })
 })