Merge pull request #2307 from broadinstitute/jb-data-types-search

bistline · web-flow · commit 061829d3b8ea · 2025-10-06T14:41:37.000-04:00
Adding search options for data types (SCP-6051)
diff --git a/app/controllers/api/v1/search_controller.rb b/app/controllers/api/v1/search_controller.rb
@@ -282,9 +282,14 @@ def index
           sort_type = params[:order].to_sym
         end
 
+        result_types = params[:data_types]&.split(',')&.map(&:to_sym) || []
+        @studies = StudySearchService.filter_results_by_data_type(@studies, result_types)
+
         # convert to array to allow appending external search results (Azul, TDR, etc.)
         @studies = @studies.to_a
 
+        # filter results by file type, if requested
+
         # perform Azul search if there are facets/terms provided by user, and they requested HCA results
         # run this before inferred search so that they are weighted and sorted correctly
         # skip if user is searching inside a collection or they are performing global gene search
@@ -391,6 +396,7 @@ def index
               # only show results where we found a hit in gene search
               @inferred_studies = Study.where(:id.in => new_genes[:study_ids])
             end
+            @inferred_studies = StudySearchService.filter_results_by_data_type(@inferred_studies, result_types)
             @inferred_accessions = @inferred_studies.pluck(:accession)
             logger.info "Found #{@inferred_accessions.count} inferred matches: #{@inferred_accessions}"
             @matching_accessions += @inferred_accessions
@@ -399,7 +405,6 @@ def index
         end
 
         @matching_accessions = @studies.map { |study| self.class.get_study_attribute(study, :accession) }
-        logger.info "studies_by_facet: #{@studies_by_facet}"
         logger.info "Final list of matching studies: #{@matching_accessions}"
         @results = @studies.paginate(page: params[:page], per_page: Study.per_page)
         if params[:export].present?
diff --git a/app/javascript/components/search/controls/OptionsButton.jsx b/app/javascript/components/search/controls/OptionsButton.jsx
@@ -10,19 +10,24 @@ export default function OptionsButton() {
   const searchContext = useContext(StudySearchContext)
   const [showOptions, setShowOptions] = useState(false)
   const configuredOptions = [
-    { searchProp: 'external', value: 'hca', label: 'Include HCA results' }
+    { searchProp: 'external', value: 'hca', label: 'Include HCA results' },
+    { searchProp: 'data_types', value: 'raw_counts', label: 'Has raw counts', multiple: true },
+    { searchProp: 'data_types', value: 'diff_exp', label: 'Has differential expression', multiple: true },
+    { searchProp: 'data_types', value: 'spatial', label: 'Has spatial data', multiple: true }
   ]
 
   const optionsPopover = <Popover data-analytics-name='search-options-menu' id='search-options-menu'>
     <ul className="facet-filter-list">
       {
-        configuredOptions.map((option) => {
+        configuredOptions.map((option, index) => {
         return <OptionsControl
-          key={option.searchProp}
+          key={`${option.searchProp}-${index}`}
           searchContext={searchContext}
           searchProp={option.searchProp}
           value={option.value}
-          label={option.label}/>
+          label={option.label}
+          multiple={option.multiple}
+        />
         })
       }
     </ul>
diff --git a/app/javascript/components/search/controls/OptionsControl.jsx b/app/javascript/components/search/controls/OptionsControl.jsx
@@ -1,13 +1,21 @@
 import React, { useState } from 'react'
 
-export default function OptionsControl({searchContext, searchProp, value, label}) {
+/** checkbox control for adding optional parameters to search query */
+export default function OptionsControl({ searchContext, searchProp, value, label, multiple = false }) {
   const defaultChecked = searchContext.params[searchProp] === value
   const [isChecked, setIsChecked] = useState(defaultChecked)
 
+
   /** toggle state of checkbox */
   function toggleCheckbox(checked) {
     setIsChecked(checked)
-    searchContext.updateSearch({ [searchProp] : checked ? value : null })
+    if (multiple) {
+      const existingOpts = searchContext.params[searchProp]?.split(',').filter(o => o !== '') || []
+      const newOpts = checked ? existingOpts.concat(value) : existingOpts.filter(v => v !== value)
+      searchContext.updateSearch({ [searchProp] : newOpts.join(',') })
+    } else {
+      searchContext.updateSearch({ [searchProp] : checked ? value : null })
+    }
   }
 
   return (
diff --git a/app/javascript/lib/scp-api.jsx b/app/javascript/lib/scp-api.jsx
@@ -926,7 +926,7 @@ export async function fetchSearch(type, searchParams, mock=false) {
 export function buildSearchQueryString(type, searchParams) {
   const facetsParam = buildFacetQueryString(searchParams.facets)
 
-  const params = ['page', 'order', 'terms', 'external', 'export', 'preset', 'genes', 'genePage']
+  const params = ['page', 'order', 'terms', 'external', 'export', 'data_types', 'preset', 'genes', 'genePage']
   let otherParamString = params.map(param => {
     return searchParams[param] ? `&${param}=${searchParams[param]}` : ''
   }).join('')
diff --git a/app/javascript/providers/GeneSearchProvider.jsx b/app/javascript/providers/GeneSearchProvider.jsx
@@ -128,6 +128,7 @@ export function buildParamsFromQuery(query, preset) {
     genes: cleanGeneParams,
     terms: queryParams.terms ? queryParams.terms : '',
     external: queryParams.external ? queryParams.external : '',
+    data_types: queryParams.data_types ? queryParams.data_types : '',
     facets: buildFacetsFromQueryString(queryParams.facets),
     preset: preset ? preset : queryString.preset_search
   }
diff --git a/app/javascript/providers/SearchSelectionProvider.jsx b/app/javascript/providers/SearchSelectionProvider.jsx
@@ -43,7 +43,7 @@ export default function SearchSelectionProvider(props) {
   const [selection, setSelection] = useState(
     appliedSelection ?
       appliedSelection :
-      { terms: '', facets: {}, external: '' })
+      { terms: '', facets: {}, external: '', data_types: '' })
   selection.updateSelection = updateSelection
   selection.updateFacet = updateFacet
   selection.performSearch = performSearch
diff --git a/app/javascript/providers/StudySearchProvider.jsx b/app/javascript/providers/StudySearchProvider.jsx
@@ -18,6 +18,7 @@ const emptySearch = {
     terms: '',
     facets: {},
     external: '',
+    data_types: '',
     page: 1,
     preset_search: undefined,
     order: undefined
@@ -75,12 +76,12 @@ export function useContextStudySearch() {
   return useContext(StudySearchContext)
 }
 
-/** Merges the external parameter into the searchParams object */
-export function mergeExternalParam(searchParams, newParams) {
-  if (Object.keys(newParams).length === 1 && Object.keys(newParams)[0] === 'external') {
-    return newParams.external
+/** Merges any optional parameter into the searchParams object */
+export function mergeOptionalParam(searchParams, newParams, paramName) {
+  if (Object.keys(newParams).length === 1 && Object.keys(newParams)[0] === paramName) {
+    return newParams[paramName]
   } else {
-    return searchParams.external
+    return searchParams[paramName]
   }
 }
 
@@ -107,7 +108,8 @@ export function PropsStudySearchProvider(props) {
     // reset the page to 1 for new searches, unless otherwise specified
     search.page = newParams.page ? newParams.page : 1
     search.preset = undefined // for now, exclude preset from the page URL--it's in the component props instead
-    search.external = mergeExternalParam(searchParams, newParams)
+    search.external = mergeOptionalParam(searchParams, newParams, 'external')
+    search.data_types = mergeOptionalParam(searchParams, newParams, 'data_types')
     const mergedParams = Object.assign(buildGeneParamsFromQuery(window.location.search), search)
     const queryString = buildSearchQueryString('study', mergedParams)
     navigate(`?${queryString}`)
@@ -165,6 +167,7 @@ export function buildParamsFromQuery(query, preset) {
     terms: queryParams.terms ? queryParams.terms : '',
     facets: buildFacetsFromQueryString(queryParams.facets),
     external: queryParams.external ? queryParams.external : '',
+    data_types: queryParams.data_types ? queryParams.data_types : '',
     preset: preset ? preset : queryString.preset_search,
     order: queryParams.order
   }
diff --git a/app/lib/study_search_service.rb b/app/lib/study_search_service.rb
@@ -199,6 +199,30 @@ def self.perform_mongo_facet_search(facets)
     end
   end
 
+  def self.filter_results_by_data_type(studies, data_types)
+    return studies if data_types.empty?
+
+    matches = data_types.index_with { [] }
+    # note: this has to be updated if a new data_type is added
+    matchers = {
+      raw_counts: :has_raw_counts_matrices?,
+      diff_exp: :has_differential_expression_results?,
+      spatial: :has_spatial_clustering?
+    }
+
+    # run matching in parallel to reduce UI blocking
+    Parallel.map(studies, in_threads: 10) do |study|
+      data_types.each do |data_type|
+        matches[data_type] << study.accession if study.send(matchers[data_type])
+      end
+    end
+
+    # find the intersection of all matches by data types
+    study_matches = matches.values
+    accessions = study_matches[0].intersection(*study_matches[1..])
+    studies.where(:accession.in => accessions)
+  end
+
   # deal with ontology id formatting inconsistencies
   def self.convert_id_format(id)
     parts = id.split(/[_:]/)
diff --git a/app/models/hca_azul_client.rb b/app/models/hca_azul_client.rb
@@ -23,7 +23,6 @@ class HcaAzulClient
   # Default headers for API requests
   DEFAULT_HEADERS = {
     'Accept' => 'application/json',
-    'Content-Type' => 'application/json',
     'x-app-id' => 'single-cell-portal',
     'x-domain-id' => "#{ENV['HOSTNAME']}"
   }.freeze
@@ -104,11 +103,23 @@ def process_api_request(http_method, path, payload: nil, retry_count: 0)
   # * *raises*
   #   - (RestClient::Exception) => if HTTP request fails for any reason
   def execute_http_request(http_method, path, payload = nil)
-    response = RestClient::Request.execute(method: http_method, url: path, payload:, headers: DEFAULT_HEADERS)
+    response = RestClient::Request.execute(method: http_method, url: path, payload:, headers: set_headers(http_method))
     # handle response using helper
     handle_response(response)
   end
 
+  # set HTTP headers based on method
+  # GET requests do not support the Content-Type header, but all PUT/POST/PATCH requests do
+  #
+  # * *params*
+  #   - +http_method+ (String, Symbol) => HTTP method, e.g. :get, :post
+  #
+  # * *returns*
+  #   - (Hash) => HTTP headers object
+  def set_headers(http_method)
+    http_method.to_sym == :get ? DEFAULT_HEADERS : DEFAULT_HEADERS.merge('Content-Type' => 'application/json')
+  end
+
   # FROM SCP-4592: Temporarily disable automatic retries while we investigate the rise in 503 errors from Azul
   def should_retry?(code)
     false
diff --git a/app/models/study.rb b/app/models/study.rb
@@ -1062,6 +1062,14 @@ def has_visualization_matrices?
     end.any?
   end
 
+  def has_differential_expression_results?
+    differential_expression_results.any?
+  end
+
+  def has_spatial_clustering?
+    spatial_cluster_groups.any?
+  end
+
   # check if study has any files that can be streamed from the bucket for visualization
   # this includes BAM, BED, inferCNV Ideogram annotations, Image files, and DE files
   #
diff --git a/test/api/search_controller_test.rb b/test/api/search_controller_test.rb
@@ -226,7 +226,7 @@ class SearchControllerTest < ActionDispatch::IntegrationTest
                  "Did not return correct array of matching accessions, expected #{expected_accessions} but found #{matching_accessions}"
 
     assert_equal @random_seed, json['studies'].first['term_matches'].first
-    assert_equal 2, json['match_by_data']['numResults:scp:text']
+    assert_equal expected_accessions.count, json['match_by_data']['numResults:scp:text']
 
     # test exact phrase
     search_phrase = '"API Test Study"'
@@ -559,4 +559,57 @@ class SearchControllerTest < ActionDispatch::IntegrationTest
     found_accessions = lines.map { |l| l.split(/\t/)[1] }.flatten.uniq.sort
     assert_equal all_accessions, found_accessions
   end
+
+  test 'should find studies based on available data types' do
+    search_study = FactoryBot.create(:detached_study,
+                                     name_prefix: "Raw Counts Search Study #{@random_seed}",
+                                     public: true,
+                                     user: @user,
+                                     test_array: @@studies_to_clean)
+    detail = search_study.build_study_detail
+    detail.full_description = '<p>This is the description.</p>'
+    detail.save!
+    cells = %w[cellA cellB cellC cellD cellE cellF cellG]
+    FactoryBot.create(
+      :metadata_file, name: 'metadata.txt', study: search_study, cell_input: cells,
+      annotation_input: [{ name: 'species', type: 'group', values: %w[dog cat dog dog cat cat cat] }]
+    )
+    coordinates = 1.upto(7).to_a
+    cluster_file = FactoryBot.create(
+      :cluster_file, name: 'cluster_diffexp.txt', study: search_study,
+      cell_input: { x: coordinates, y: coordinates, cells: }
+    )
+
+    %w[raw_counts diff_exp spatial].each do |data_type|
+      execute_http_request(:get, api_v1_search_path(type: 'study', data_types: data_type))
+      assert_response :success
+      assert_empty json['studies']
+    end
+
+    # create raw counts matrix
+    exp_matrix = FactoryBot.create(:expression_file,
+                                   name: 'matrix.tsv',
+                                   study: search_study)
+    exp_matrix.build_expression_file_info(
+      is_raw_counts: true, units: 'raw counts', library_preparation_protocol: "10x 5' v3",
+      modality: 'Transcriptomic: unbiased', biosample_input_type: 'Whole cell'
+    )
+    exp_matrix.save
+    # create DE results
+    DifferentialExpressionResult.create(
+      study: search_study, cluster_group: cluster_file.cluster_groups.first, annotation_name: 'species',
+      annotation_scope: 'study', matrix_file_id: exp_matrix.id
+    )
+    # create spatial cluster
+    FactoryBot.create(
+      :cluster_file, name: 'spatial.txt', study: search_study, is_spatial: true,
+      cell_input: { x: coordinates, y: coordinates, cells: }
+    )
+
+    %w[raw_counts diff_exp spatial].each do |data_type|
+      execute_http_request(:get, api_v1_search_path(type: 'study', data_types: data_type))
+      assert_response :success
+      assert_equal search_study.accession, json['studies'].first['accession']
+    end
+  end
 end
diff --git a/test/js/search/options-control.test.js b/test/js/search/options-control.test.js
@@ -38,4 +38,24 @@ describe('OptionsControl component', () => {
     expect(checkbox).toBeChecked()
     expect(searchContext.updateSearch).toHaveBeenCalledWith({ 'external': 'hca' })
   })
+
+  it('merges multiple option controls into same parameter', () => {
+    const searchContext = {
+      params: { 'data_types': 'raw_counts' },
+      updateSearch: jest.fn()
+    }
+    const { getByText } = render(
+      <>
+        <OptionsControl
+          searchContext={searchContext} searchProp='data_types' value='raw_counts'
+          label='Has raw counts' multiple={true}
+        />
+        <OptionsControl
+          searchContext={searchContext} searchProp='data_types' value='spatial' label='Has spatial' multiple={true}
+        />
+      </>
+    )
+    fireEvent.click(getByText('Has spatial'))
+    expect(searchContext.updateSearch).toHaveBeenCalledWith({ 'data_types': 'raw_counts,spatial' })
+  })
 })

Original file line number	Diff line number	Diff line change
`@@ -128,6 +128,7 @@ export function buildParamsFromQuery(query, preset) {`
`128`	`128`	`genes: cleanGeneParams,`
`129`	`129`	`terms: queryParams.terms ? queryParams.terms : '',`
`130`	`130`	`external: queryParams.external ? queryParams.external : '',`
	`131`	`+ data_types: queryParams.data_types ? queryParams.data_types : '',`
`131`	`132`	`facets: buildFacetsFromQueryString(queryParams.facets),`
`132`	`133`	`preset: preset ? preset : queryString.preset_search`
`133`	`134`	`}`