Merge branch 'features/v3.0.0/gohan-gene-api' into releases/v3.0.0

.gitignore

@@ -46,4 +46,7 @@ bin/*
 */*/tmp
 
 *.vcf
-*.vcf.gz
+*.vcf.gz
+
+*/*/*.csv
+*/*/*.gtf*

src/api/main.go

@@ -124,27 +124,27 @@ func main() {
 
 	e.GET("/variants/get/by/variantId", mvc.VariantsGetByVariantId,
 		// middleware
-		gam.MandateChromosomeAttribute,
+		gam.ValidateOptionalChromosomeAttribute,
 		gam.MandateCalibratedBounds,
 		gam.MandateAssemblyIdAttribute,
 		gam.ValidatePotentialGenotypeQueryParameter)
 	e.GET("/variants/get/by/sampleId", mvc.VariantsGetBySampleId,
 		// middleware
-		gam.MandateChromosomeAttribute,
+		gam.ValidateOptionalChromosomeAttribute,
 		gam.MandateCalibratedBounds,
 		gam.MandateAssemblyIdAttribute,
 		gam.MandateSampleIdsPluralAttribute,
 		gam.ValidatePotentialGenotypeQueryParameter)
 
 	e.GET("/variants/count/by/variantId", mvc.VariantsCountByVariantId,
 		// middleware
-		gam.MandateChromosomeAttribute,
+		gam.ValidateOptionalChromosomeAttribute,
 		gam.MandateCalibratedBounds,
 		gam.MandateAssemblyIdAttribute,
 		gam.ValidatePotentialGenotypeQueryParameter)
 	e.GET("/variants/count/by/sampleId", mvc.VariantsCountBySampleId,
 		// middleware
-		gam.MandateChromosomeAttribute,
+		gam.ValidateOptionalChromosomeAttribute,
 		gam.MandateCalibratedBounds,
 		gam.MandateAssemblyIdAttribute,
 		gam.MandateSampleIdsSingularAttribute,
@@ -155,6 +155,12 @@ func main() {
 		gam.MandateAssemblyIdAttribute)
 	e.GET("/variants/ingestion/requests", mvc.GetAllVariantIngestionRequests)
 
+	// -- Genes
+	e.GET("/genes/overview", mvc.GetGenesOverview)
+	e.GET("/genes/search", mvc.GenesGetByNomenclatureWildcard,
+		// middleware
+		gam.ValidateOptionalChromosomeAttribute)
+
 	// Run
 	e.Logger.Fatal(e.Start(":" + cfg.Api.Port))
 }

src/api/middleware/chromosomeMiddleware.go

@@ -1,34 +1,25 @@
 package middleware
 
 import (
+	"api/models/constants/chromosome"
 	"net/http"
-	"strconv"
 
 	"github.com/labstack/echo"
 )
 
 /*
 	Echo middleware to ensure a valid `chromosome` HTTP query parameter was provided
 */
-func MandateChromosomeAttribute(next echo.HandlerFunc) echo.HandlerFunc {
+func ValidateOptionalChromosomeAttribute(next echo.HandlerFunc) echo.HandlerFunc {
 	return func(c echo.Context) error {
 		// check for chromosome query parameter
 		chromQP := c.QueryParam("chromosome")
-		if len(chromQP) == 0 {
-			// if no id was provided return an error
-			return echo.NewHTTPError(http.StatusBadRequest, "Missing 'chromosome' query parameter for querying!")
-		}
 
 		// verify:
-		i, conversionErr := strconv.Atoi(chromQP)
-		if conversionErr != nil {
-			// if invalid chromosome
-			return echo.NewHTTPError(http.StatusBadRequest, "Error converting 'chromosome' query parameter! Check your input")
-		}
-
-		if i <= 0 {
-			// if chromosome less than 0
-			return echo.NewHTTPError(http.StatusBadRequest, "Please provide a 'chromosome' greater than 0!")
+		if len(chromQP) > 0 && !chromosome.IsValidHumanChromosome(chromQP) {
+			// if chromosome less than 1 or greater than 23
+			// and not 'x', 'y' or 'm'
+			return echo.NewHTTPError(http.StatusBadRequest, "Please provide a valid 'chromosome' (either 1-23, X, Y, or M)")
 		}
 
 		return next(c)

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

@@ -11,6 +11,8 @@ const (
 	GRCh38 constants.AssemblyId = "GRCh38"
 	GRCh37 constants.AssemblyId = "GRCh37"
 	NCBI36 constants.AssemblyId = "NCBI36"
+	NCBI35 constants.AssemblyId = "NCBI35"
+	NCBI34 constants.AssemblyId = "NCBI34"
 	Other  constants.AssemblyId = "Other"
 )
 
@@ -22,6 +24,10 @@ func CastToAssemblyId(text string) constants.AssemblyId {
 		return GRCh37
 	case "ncbi36":
 		return NCBI36
+	case "ncbi35":
+		return NCBI35
+	case "ncbi34":
+		return NCBI34
 	case "other":
 		return Other
 	default:

src/api/models/constants/chromosome/main.go

@@ -0,0 +1,37 @@
+package chromosome
+
+import (
+	"strconv"
+	"strings"
+)
+
+func IsValidHumanChromosome(text string) bool {
+
+	// Check if number can be represented as an int as is non-zero
+	chromNumber, _ := strconv.Atoi(text)
+	if chromNumber > 0 {
+		// It can..
+		// Check if it in range 1-23
+		if chromNumber < 24 {
+			return true
+		}
+	} else {
+		// No it can't..
+		// Check if it is an X, Y..
+		loweredText := strings.ToLower(text)
+		switch loweredText {
+		case "x":
+			return true
+		case "y":
+			return true
+		}
+
+		// ..or M (MT)
+		switch strings.Contains(loweredText, "m") {
+		case true:
+			return true
+		}
+	}
+
+	return false
+}

src/api/models/constants/main.go

@@ -7,6 +7,7 @@ package constants
 	associated services.
 */
 type AssemblyId string
+type Chromosome string
 type GenotypeQuery string
 type SearchOperation string
 type SortDirection string

src/api/models/dtos.go

@@ -12,3 +12,11 @@ type VariantResponseDataModel struct {
 	Count     int       `json:"count"`
 	Results   []Variant `json:"results"` // []Variant
 }
+
+type GenesResponseDTO struct {
+	Status  int    `json:"status"`
+	Message string `json:"message"`
+	Term    string `json:"term"`
+	Count   int    `json:"count"`
+	Results []Gene `json:"results"` // []Gene
+}

src/api/models/elasticsearch.go

@@ -7,7 +7,7 @@ import (
 var VcfHeaders = []string{"chrom", "pos", "id", "ref", "alt", "qual", "filter", "info", "format"}
 
 type Variant struct {
-	Chrom  int      `json:"chrom"`
+	Chrom  string   `json:"chrom"`
 	Pos    int      `json:"pos"`
 	Id     string   `json:"id"`
 	Ref    []string `json:"ref"`
@@ -45,3 +45,11 @@ type Genotype struct {
 	AlleleRight int        `json:"alleleRight"` // -1 = no call (equivalent to a '.')
 	Zygosity    c.Zygosity `json:"zygosity"`
 }
+
+type Gene struct {
+	Name       string       `json:"name"`
+	Chrom      string       `json:"chrom"`
+	Start      int          `json:"start"`
+	End        int          `json:"end"`
+	AssemblyId c.AssemblyId `json:"assemblyId"`
+}

src/api/models/ingest/structs/main.go

@@ -0,0 +1,16 @@
+package structs
+
+import (
+	"api/models"
+	"sync"
+)
+
+type IngestionQueueStructure struct {
+	Variant   *models.Variant
+	WaitGroup *sync.WaitGroup
+}
+
+type GeneIngestionQueueStructure struct {
+	Gene      *models.Gene
+	WaitGroup *sync.WaitGroup
+}

src/api/mvc/genes.go

@@ -0,0 +1,152 @@
+package mvc
+
+import (
+	"api/contexts"
+	"api/models"
+	assemblyId "api/models/constants/assembly-id"
+	esRepo "api/repositories/elasticsearch"
+	"fmt"
+	"net/http"
+	"strconv"
+	"sync"
+
+	"github.com/labstack/echo"
+	"github.com/mitchellh/mapstructure"
+)
+
+func GenesGetByNomenclatureWildcard(c echo.Context) error {
+	cfg := c.(*contexts.GohanContext).Config
+	es := c.(*contexts.GohanContext).Es7Client
+
+	// Chromosome search term
+	chromosomeSearchTerm := c.QueryParam("chromosome")
+	if len(chromosomeSearchTerm) == 0 {
+		// if no chromosome is provided, assume "wildcard" search
+		chromosomeSearchTerm = "*"
+	}
+
+	// Name search term
+	term := c.QueryParam("term")
+
+	// Assembly ID
+	// perform wildcard search if empty/random parameter is passed
+	// - set to Unknown to trigger it
+	assId := assemblyId.Unknown
+	if assemblyId.CastToAssemblyId(c.QueryParam("assemblyId")) != assemblyId.Unknown {
+		// retrieve passed parameter if is valid
+		assId = assemblyId.CastToAssemblyId(c.QueryParam("assemblyId"))
+	}
+
+	// Size
+	var (
+		size        int = 25
+		sizeCastErr error
+	)
+	if len(c.QueryParam("size")) > 0 {
+		sizeQP := c.QueryParam("size")
+		size, sizeCastErr = strconv.Atoi(sizeQP)
+		if sizeCastErr != nil {
+			size = 25
+		}
+	}
+
+	fmt.Printf("Executing wildcard genes search for term %s, assemblyId %s (max size: %d)\n", term, assId, size)
+
+	// Execute
+	docs := esRepo.GetGeneDocumentsByTermWildcard(cfg, es, chromosomeSearchTerm, term, assId, size)
+
+	docsHits := docs["hits"].(map[string]interface{})["hits"]
+	allDocHits := []map[string]interface{}{}
+	mapstructure.Decode(docsHits, &allDocHits)
+
+	// grab _source for each hit
+	var allSources []models.Gene
+
+	for _, r := range allDocHits {
+		source := r["_source"].(map[string]interface{})
+
+		// cast map[string]interface{} to struct
+		var resultingVariant models.Gene
+		mapstructure.Decode(source, &resultingVariant)
+
+		// accumulate structs
+		allSources = append(allSources, resultingVariant)
+	}
+
+	fmt.Printf("Found %d docs!\n", len(allSources))
+
+	geneResponseDTO := models.GenesResponseDTO{
+		Term:    term,
+		Count:   len(allSources),
+		Results: allSources,
+		Status:  200,
+		Message: "Success",
+	}
+
+	return c.JSON(http.StatusOK, geneResponseDTO)
+}
+
+func GetGenesOverview(c echo.Context) error {
+
+	resultsMap := map[string]interface{}{}
+	resultsMux := sync.RWMutex{}
+
+	es := c.(*contexts.GohanContext).Es7Client
+	cfg := c.(*contexts.GohanContext).Config
+
+	// retrieve aggregation of genes/chromosomes by assembly id
+	results := esRepo.GetGeneBucketsByKeyword(cfg, es)
+
+	// begin mapping results
+	geneChromosomeGroupBucketsMapped := []map[string]interface{}{}
+
+	// loop over top level aggregation and
+	// accumulated nested aggregations
+	if aggs, ok := results["aggregations"]; ok {
+		aggsMapped := aggs.(map[string]interface{})
+
+		if items, ok := aggsMapped["genes_assembly_id_group"]; ok {
+			itemsMapped := items.(map[string]interface{})
+
+			if buckets := itemsMapped["buckets"]; ok {
+				arrayMappedBuckets := buckets.([]interface{})
+
+				for _, mappedBucket := range arrayMappedBuckets {
+					geneChromosomeGroupBucketsMapped = append(geneChromosomeGroupBucketsMapped, mappedBucket.(map[string]interface{}))
+				}
+			}
+		}
+	}
+
+	individualAssemblyIdKeyMap := map[string]interface{}{}
+
+	// iterated over each assemblyId bucket
+	for _, chromGroupBucketMap := range geneChromosomeGroupBucketsMapped {
+
+		assemblyIdKey := fmt.Sprint(chromGroupBucketMap["key"])
+
+		numGenesPerChromMap := map[string]interface{}{}
+		bucketsMapped := map[string]interface{}{}
+
+		if chromGroupItem, ok := chromGroupBucketMap["genes_chromosome_group"]; ok {
+			chromGroupItemMapped := chromGroupItem.(map[string]interface{})
+
+			for _, chromBucket := range chromGroupItemMapped["buckets"].([]interface{}) {
+				doc_key := fmt.Sprint(chromBucket.(map[string]interface{})["key"]) // ensure strings and numbers are expressed as strings
+				doc_count := chromBucket.(map[string]interface{})["doc_count"]
+
+				// add to list of buckets by chromosome
+				bucketsMapped[doc_key] = doc_count
+			}
+		}
+
+		numGenesPerChromMap["numberOfGenesPerChromosome"] = bucketsMapped
+		individualAssemblyIdKeyMap[assemblyIdKey] = numGenesPerChromMap
+	}
+
+	resultsMux.Lock()
+	resultsMap["assemblyIDs"] = individualAssemblyIdKeyMap
+	resultsMux.Unlock()
+
+	return c.JSON(http.StatusOK, resultsMap)
+}

src/api/mvc/variants.go

@@ -249,7 +249,7 @@ func GetVariantsOverview(c echo.Context) error {
 	callGetBucketsByKeyword := func(key string, keyword string, _wg *sync.WaitGroup) {
 		defer _wg.Done()
 
-		results := esRepo.GetBucketsByKeyword(cfg, es, keyword)
+		results := esRepo.GetVariantsBucketsByKeyword(cfg, es, keyword)
 
 		// retrieve aggregations.items.buckets
 		bucketsMapped := []interface{}{}
@@ -281,7 +281,7 @@ func GetVariantsOverview(c echo.Context) error {
 
 	// get distribution of chromosomes
 	wg.Add(1)
-	go callGetBucketsByKeyword("chromosomes", "chrom", &wg)
+	go callGetBucketsByKeyword("chromosomes", "chrom.keyword", &wg)
 
 	// get distribution of variant IDs
 	wg.Add(1)