feat!: use reference service to provide assembly ID

src/api/contexts/contexts.go

@@ -25,7 +25,7 @@ type (
 
 	// Convenient storage for relevant http context data
 	QueryParameters struct {
-		AssemblyId constants.AssemblyId
+		AssemblyId string
 		Alleles    []string
 		Chromosome string
 		Genotype   constants.GenotypeQuery

src/api/middleware/assemblyMiddleware.go

@@ -2,8 +2,6 @@ package middleware
 
 import (
 	"gohan/api/contexts"
-	"gohan/api/models/constants"
-	assid "gohan/api/models/constants/assembly-id"
 	"net/http"
 
 	"github.com/labstack/echo"
@@ -16,14 +14,14 @@ func MandateAssemblyIdAttribute(next echo.HandlerFunc) echo.HandlerFunc {
 	return func(c echo.Context) error {
 		// check for assemblyId query parameter
 		assemblyId := c.QueryParam("assemblyId")
-		if len(assemblyId) == 0 || !assid.IsKnownAssemblyId(assemblyId) {
-			// if no id was provided, or it was invalid, return an error
-			return echo.NewHTTPError(http.StatusBadRequest, "Missing or unknown assemblyId!")
+		if len(assemblyId) == 0 {
+			// if no id was provided, return an error
+			return echo.NewHTTPError(http.StatusBadRequest, "Missing assemblyId!")
 		}
 
 		// forward a type-safe value down the pipeline
 		gc := c.(*contexts.GohanContext)
-		gc.AssemblyId = constants.AssemblyId(assemblyId)
+		gc.AssemblyId = assemblyId
 
 		return next(gc)
 	}

src/api/models/constants/assembly-id/main.go

@@ -2,41 +2,9 @@ package assemblyId
 
 import (
 	"gohan/api/models/constants"
-	"strings"
 )
 
 const (
-	Unknown constants.AssemblyId = "Unknown"
-
 	GRCh38 constants.AssemblyId = "GRCh38"
 	GRCh37 constants.AssemblyId = "GRCh37"
-	NCBI36 constants.AssemblyId = "NCBI36"
-	NCBI35 constants.AssemblyId = "NCBI35"
-	NCBI34 constants.AssemblyId = "NCBI34"
-	Other  constants.AssemblyId = "Other"
 )
-
-func CastToAssemblyId(text string) constants.AssemblyId {
-	switch strings.ToLower(text) {
-	case "grch38":
-		return GRCh38
-	case "grch37":
-		return GRCh37
-	case "ncbi36":
-		return NCBI36
-	case "ncbi35":
-		return NCBI35
-	case "ncbi34":
-		return NCBI34
-	case "other":
-		return Other
-	default:
-		return Unknown
-	}
-}
-
-func IsKnownAssemblyId(text string) bool {
-	// attempt to cast to assemblyId and
-	// return if unknown assemblyId
-	return CastToAssemblyId(text) != Unknown
-}

src/api/mvc/genes/main.go

@@ -335,11 +335,7 @@ func GenesGetByNomenclatureWildcard(c echo.Context) error {
 	// Assembly ID
 	// perform wildcard search if empty/random parameter is passed
 	// - set to Unknown to trigger it
-	var assId constants.AssemblyId
-	if gc.AssemblyId != assemblyId.Unknown {
-		// retrieve passed parameter if is valid
-		assId = gc.AssemblyId
-	}
+	asmId := gc.AssemblyId
 
 	// Size
 	var (
@@ -354,10 +350,10 @@ func GenesGetByNomenclatureWildcard(c echo.Context) error {
 		}
 	}
 
-	fmt.Printf("Executing wildcard genes search for term %s, assemblyId %s (max size: %d)\n", term, assId, size)
+	fmt.Printf("Executing wildcard genes search for term %s, assemblyId %s (max size: %d)\n", term, asmId, size)
 
 	// Execute
-	docs, geneErr := esRepo.GetGeneDocumentsByTermWildcard(cfg, es, chromosomeSearchTerm, term, assId, size)
+	docs, geneErr := esRepo.GetGeneDocumentsByTermWildcard(cfg, es, chromosomeSearchTerm, term, asmId, size)
 	if geneErr != nil {
 		return c.JSON(http.StatusOK, map[string]interface{}{
 			"status":  500,

src/api/mvc/main.go

@@ -3,7 +3,6 @@ package mvc
 import (
 	"gohan/api/contexts"
 	"gohan/api/models/constants"
-	a "gohan/api/models/constants/assembly-id"
 	gq "gohan/api/models/constants/genotype-query"
 	"strings"
 
@@ -12,7 +11,7 @@ import (
 	"github.com/labstack/echo"
 )
 
-func RetrieveCommonElements(c echo.Context) (*elasticsearch.Client, string, int, int, string, string, []string, constants.GenotypeQuery, constants.AssemblyId, string) {
+func RetrieveCommonElements(c echo.Context) (*elasticsearch.Client, string, int, int, string, string, []string, constants.GenotypeQuery, string, string) {
 	gc := c.(*contexts.GohanContext)
 	es := gc.Es7Client
 
@@ -48,11 +47,7 @@ func RetrieveCommonElements(c echo.Context) (*elasticsearch.Client, string, int,
 		}
 	}
 
-	assemblyId := a.Unknown
-	assemblyIdQP := c.QueryParam("assemblyId")
-	if len(assemblyIdQP) > 0 && a.IsKnownAssemblyId(assemblyIdQP) {
-		assemblyId = a.CastToAssemblyId(assemblyIdQP)
-	}
+	assemblyId := c.QueryParam("assemblyId")
 
 	return es, chromosome, lowerBound, upperBound, reference, alternative, alleles, genotype, assemblyId, datasetString
 }

src/api/repositories/elasticsearch/genes.go

@@ -13,7 +13,6 @@ import (
 
 	"gohan/api/models"
 	"gohan/api/models/constants"
-	assemblyId "gohan/api/models/constants/assembly-id"
 	"gohan/api/utils"
 
 	"github.com/elastic/go-elasticsearch/v7"
@@ -106,7 +105,7 @@ func GetGeneBucketsByKeyword(cfg *models.Config, es *elasticsearch.Client) (map[
 }
 
 func GetGeneDocumentsByTermWildcard(cfg *models.Config, es *elasticsearch.Client,
-	chromosomeSearchTerm string, term string, assId constants.AssemblyId, size int) (map[string]interface{}, error) {
+	chromosomeSearchTerm string, term string, asmId string, size int) (map[string]interface{}, error) {
 
 	if cfg.Debug {
 		http.DefaultTransport.(*http.Transport).TLSClientConfig = &tls.Config{InsecureSkipVerify: true}
@@ -115,10 +114,10 @@ func GetGeneDocumentsByTermWildcard(cfg *models.Config, es *elasticsearch.Client
 	// Nomenclature Search Term
 	nomenclatureStringTerm := fmt.Sprintf("*%s*", term)
 
-	// Assembly Id Search Term (wildcard by default)
+	// Assembly ID Search Term (wildcard by default)
 	assemblyIdStringTerm := "*"
-	if assId != assemblyId.Unknown {
-		assemblyIdStringTerm = string(assId)
+	if asmId != "" {
+		assemblyIdStringTerm = asmId
 	}
 
 	var buf bytes.Buffer

src/api/repositories/elasticsearch/variants.go

@@ -13,7 +13,6 @@ import (
 
 	"gohan/api/models"
 	c "gohan/api/models/constants"
-	a "gohan/api/models/constants/assembly-id"
 	gq "gohan/api/models/constants/genotype-query"
 	s "gohan/api/models/constants/sort"
 	z "gohan/api/models/constants/zygosity"
@@ -465,7 +464,7 @@ func CountDocumentsContainerVariantOrSampleIdInPositionRange(cfg *models.Config,
 			}})
 	}
 
-	if assemblyId != "" && assemblyId != a.Unknown {
+	if assemblyId != "" {
 		mustMap = append(mustMap, map[string]interface{}{
 			"match": map[string]interface{}{
 				"assemblyId": map[string]interface{}{

src/api/tests/build/api/genes_test.go

@@ -6,8 +6,6 @@ import (
 	common "gohan/api/tests/common"
 
 	"gohan/api/models"
-	c "gohan/api/models/constants"
-	a "gohan/api/models/constants/assembly-id"
 	ingest "gohan/api/models/ingest"
 
 	"gohan/api/models/constants/chromosome"
@@ -205,10 +203,10 @@ func getAllDtosOfVariousCombinationsOfGenesAndAssemblyIDs(_t *testing.T) []dtos.
 				go func(_wg *sync.WaitGroup, _assemblyIdString string, _chromosomeString string) {
 					defer _wg.Done()
 
-					assemblyId := a.CastToAssemblyId(_assemblyIdString)
+					assemblyId := _assemblyIdString
 
 					// make the call
-					dto := buildQueryAndMakeGetGenesCall(_chromosomeString, "", assemblyId, _t, cfg)
+					dto := buildQueryAndMakeGetGenesCall(_chromosomeString, assemblyId, _t, cfg)
 
 					// ensure there is data returned
 					// (we'd be making a bad query, otherwise)
@@ -231,7 +229,7 @@ func getAllDtosOfVariousCombinationsOfGenesAndAssemblyIDs(_t *testing.T) []dtos.
 	return allDtoResponses
 }
 
-func buildQueryAndMakeGetGenesCall(chromosome string, term string, assemblyId c.AssemblyId, _t *testing.T, _cfg *models.Config) dtos.GenesResponseDTO {
+func buildQueryAndMakeGetGenesCall(chromosome string, assemblyId string, _t *testing.T, _cfg *models.Config) dtos.GenesResponseDTO {
 
 	queryString := fmt.Sprintf("?chromosome=%s&assemblyId=%s", chromosome, assemblyId)
 

src/api/tests/build/api/variants_test.go

@@ -18,7 +18,6 @@ import (
 	"time"
 
 	c "gohan/api/models/constants"
-	a "gohan/api/models/constants/assembly-id"
 	gq "gohan/api/models/constants/genotype-query"
 	s "gohan/api/models/constants/sort"
 	z "gohan/api/models/constants/zygosity"
@@ -681,7 +680,7 @@ func getAllDtosOfVariousCombinationsOfChromosomesAndSampleIds(_t *testing.T, inc
 
 			chrom := _combination[0]
 			sampleId := _combination[1]
-			assemblyId := a.CastToAssemblyId(_combination[2])
+			assemblyId := _combination[2]
 
 			// make the call
 			dto := common.BuildQueryAndMakeGetVariantsCall(chrom, sampleId, uuid.Nil, includeInfo, sortByPosition, genotype, assemblyId, referenceAllelePattern, alternativeAllelePattern, "", false, _t, cfg)

src/api/tests/common/common.go

@@ -6,7 +6,6 @@ import (
 	"fmt"
 	"gohan/api/models"
 	c "gohan/api/models/constants"
-	a "gohan/api/models/constants/assembly-id"
 	gq "gohan/api/models/constants/genotype-query"
 	s "gohan/api/models/constants/sort"
 	testConsts "gohan/api/tests/common/constants"
@@ -215,7 +214,7 @@ func GetAndVerifyVariantsResults(_cfg *models.Config, _t *testing.T, dataset uui
 
 func BuildQueryAndMakeGetVariantsCall(
 	chromosome string, sampleId string, dataset uuid.UUID, includeInfo bool,
-	sortByPosition c.SortDirection, genotype c.GenotypeQuery, assemblyId c.AssemblyId,
+	sortByPosition c.SortDirection, genotype c.GenotypeQuery, assemblyId string,
 	referenceAllelePattern string, alternativeAllelePattern string, commaDeliminatedAlleles string,
 	ignoreStatusCode bool, _t *testing.T, _cfg *models.Config) dtos.VariantGetReponse {
 
@@ -286,7 +285,7 @@ func GetAllDtosOfVariousCombinationsOfChromosomesAndSampleIds(_t *testing.T, dat
 
 			chrom := _combination[0]
 			sampleId := _combination[1]
-			assemblyId := a.CastToAssemblyId(_combination[2])
+			assemblyId := _combination[2]
 
 			// make the call
 			dto := BuildQueryAndMakeGetVariantsCall(chrom, sampleId, dataset, includeInfo, sortByPosition, genotype, assemblyId, referenceAllelePattern, alternativeAllelePattern, "", false, _t, cfg)

src/api/workflows/main.go

@@ -1,10 +1,5 @@
 package workflows
 
-import (
-	c "gohan/api/models/constants"
-	a "gohan/api/models/constants/assembly-id"
-)
-
 type WorkflowSchema map[string]interface{}
 
 var WORKFLOW_VARIANT_SCHEMA WorkflowSchema = map[string]interface{}{
@@ -22,6 +17,7 @@ var WORKFLOW_VARIANT_SCHEMA WorkflowSchema = map[string]interface{}{
 					"id":       "project_dataset",
 					"type":     "project:dataset",
 					"required": true,
+					"help":     "The dataset to ingest the variants into.",
 				},
 				{
 					"id":       "vcf_gz_file_names",
@@ -33,12 +29,14 @@ var WORKFLOW_VARIANT_SCHEMA WorkflowSchema = map[string]interface{}{
 					"id":       "assembly_id",
 					"type":     "enum",
 					"required": true,
-					"values":   []c.AssemblyId{a.GRCh38, a.GRCh37},
+					"values":   "{{ serviceUrls.reference }}/genomes?response_format=id_list",
 				},
 				{
 					"id":       "filter_out_references",
 					"type":     "boolean",
 					"required": true,
+					"help": "If this is checked, variant calls which are (0, 0) (i.e., homozygous reference " +
+						"calls) will not be ingested.",
 				},
 				// Injected inputs:
 				{