add by_alt to address #68

README.md

@@ -118,6 +118,7 @@ this case, the op determines how the many values are `reduced`. Valid operations
  + min      // numbers only
  + sum      // numbers only
  + uniq     // comma-delimited list of uniq vlues
+ + by_alt   // comma-delimited by alt, pipe-delimited (|) for multiple annos for the same alt.
 
 There are some operations that are only for `postannotation`:
 

api/api.go

@@ -201,6 +201,24 @@ func allEqual(a, b []string) bool {
 	return true
 }
 
+// given the output from handleA and the alts:
+// append new values to the appropriate alt.
+// 22,33, A,G -> 22,33
+// then XX, G -> 22,33|G
+// then YY, A -> 22|YY,33|G
+func byAlt(in []interface{}, qAlts []string, existing [][]string) [][]string {
+	if existing == nil {
+		existing = make([][]string, len(qAlts))
+	}
+	for i, v := range in {
+		if v == "." || v == "" || v == nil {
+			continue
+		}
+		existing[i] = append(existing[i], fmt.Sprintf("%v", v))
+	}
+	return existing
+}
+
 // handleA converts the `val` to the correct slice of vals to match what's isnt
 // qAlts and oAlts. Then length of the returned value should always be equal
 // to the len of qAlts.
@@ -257,6 +275,7 @@ func handleA(val interface{}, qAlts []string, oAlts []string, out []interface{})
 func collect(v interfaces.IVariant, rels []interfaces.Relatable, src *Source, strict bool) ([]interface{}, error) {
 	coll := make([]interface{}, 0, len(rels))
 	var val interface{}
+	var valByAlt [][]string
 	var finalerr error
 	for _, other := range rels {
 		if int(other.Source())-1 != src.Index {
@@ -293,15 +312,17 @@ func collect(v interfaces.IVariant, rels []interfaces.Relatable, src *Source, st
 					continue
 				}
 			}
+			if src.Op == "by_alt" {
+				// with alt uses handleA machinery and then concats each value with then
+				// alternate allele.
+				out := make([]interface{}, len(v.Alt()))
+				handleA(val, v.Alt(), o.Alt(), out)
+				valByAlt = byAlt(out, v.Alt(), valByAlt)
+				continue
+			}
 
-			/*
-				if src.Field == "ID" || src.Field == "FILTER" {
-					coll = append(coll, val)
-					continue
-				}
-			*/
 			// special-case 'self' when the annotation has Number=A and either query or anno have multiple alts
-			// note that if len(rels) > 1, we could miss some since we return here. however, that shouldn't happen as we are matching on ref and alt and we wouldn't know what to do anyway.
+			// so that we get the alts matched up.
 			if src.NumberA && src.Op == "self" && src.Field != "ID" && src.Field != "FILTER" {
 				var out []interface{}
 				if len(coll) > 0 {
@@ -390,6 +411,15 @@ func collect(v interfaces.IVariant, rels []interfaces.Relatable, src *Source, st
 			coll = []interface{}{msg}
 		}
 	}
+	if valByAlt != nil {
+		for _, v := range valByAlt {
+			if len(v) == 0 {
+				coll = append(coll, ".")
+			} else {
+				coll = append(coll, strings.Join(v, "|"))
+			}
+		}
+	}
 	return coll, finalerr
 }
 
@@ -460,7 +490,10 @@ func (s *Source) AnnotateOne(v interfaces.IVariant, vals []interface{}, prefix s
 func (s *Source) UpdateHeader(r HeaderUpdater, ends bool, htype string, number string, desc string) {
 	ntype := "String"
 	// for 'self' and 'first', we can get the type from the header of the annotation file.
-	if htype != "" && (s.Op == "self" || s.Op == "first") {
+	if s.Op == "by_alt" {
+		number = "A"
+		ntype = htype
+	} else if htype != "" && (s.Op == "self" || s.Op == "first") {
 		ntype = htype
 	} else {
 		if strings.HasSuffix(s.Name, "_float") {

api/api_test.go

@@ -348,3 +348,33 @@ func TestHandlAMulti(t *testing.T) {
 	}
 
 }
+
+// given the output from handleA and the alts:
+// append new values to the appropriate alt.
+// 22,33, A,G -> 22,33
+// then XX, G -> 22,33|G
+// then YY, A -> 22|YY,33|G
+//  func byAlt(in []interface{}, qAlts []string, existing [][]string) [][]string {
+
+var byAltTests = []struct {
+	in  []interface{}
+	out [][]string
+}{
+	{[]interface{}{"AAA", "."}, [][]string{[]string{"AAA"}, nil}},
+	{[]interface{}{"AAA", "BBB"}, [][]string{[]string{"AAA"}, []string{"BBB"}}},
+	{[]interface{}{".", "BBB"}, [][]string{nil, []string{"BBB"}}},
+}
+
+func TestByAlt(t *testing.T) {
+
+	qAlts := []string{"C", "T"}
+	for _, tt := range byAltTests {
+		var existing [][]string
+
+		existing = byAlt(tt.in, qAlts, existing)
+
+		if !reflect.DeepEqual(existing, tt.out) {
+			t.Fatalf("got %v. expected %v", existing, tt.out)
+		}
+	}
+}

api/reducers.go

@@ -272,4 +272,5 @@ var Reducers = map[string]Reducer{
 	"div2":   Reducer(div2),
 	"DP2":    Reducer(dp2),
 	"setid":  Reducer(setid),
+	"by_alt": Reducer(concat),
 }

docs/CHANGES.md

@@ -1,9 +1,14 @@
 v0.2.7 (dev)
 ------------
 + restore multiple threads per annotation file after fix in biogo/hts (#64)
-+ add `setid` builtin to set the ID field and remove need for lua and fix some bugs.
++ new op `setid` builtin to set the ID field and remove need for lua and fix some bugs.
 + fix bug in `self` with multiple alternates when there were multiple overlaps (thanks Matthew).
-+ fix for #68, #69
++ fix for #68, #69 -- with Number=A, op="self", the output will always have a number of elements
+  equal to the number of alternates. If there are > 1 annos for a given site, later values will
+  overwrite previous.
++ new `op` by_alt that will have Number=A and will append multiple annotations for the same alternate
+  (from the same file) and output them as pipe-delimited. e.g with 2 alts, it might look like: `0.111,0.222|0.333`
+  when the 1st alternate has a single value and the 2nd alternate has 2 values.
 
 v0.2.6
 ------

docs/index.md

@@ -118,6 +118,7 @@ this case, the op determines how the many values are `reduced`. Valid operations
  + min      // numbers only
  + sum      // numbers only
  + uniq     // comma-delimited list of uniq vlues
+ + by_alt   // comma-delimited by alt, pipe-delimited (|) for multiple annos for the same alt.
 
 There are some operations that are only for `postannotation`:
 

tests/functional-test.sh

@@ -138,6 +138,8 @@ multiallelics() {
 }
 run check_multiallelics multiallelics
 assert_exit_code 0
+assert_in_stdout "STR_by=hello,goodbye|goodbye-again"
+assert_in_stdout "AF_by=0.000599042|0.99,0.00299521"
 
 idtest() {
     vcfanno -lua tests/id-test/some.lua tests/id-test/small.toml tests/id-test/small.vcf.gz

tests/id-test/small.toml

@@ -2,7 +2,7 @@
 file="tests/id-test/dbsnp.small.vcf.gz"
 fields=["ID", "CAF"]
 names=["rs_ids", "CAF"]
-ops=["self", "self"]
+ops=["by_alt", "self"]
 
 [[annotation]]
 file="tests/id-test/cosmic.small.vcf.gz"

tests/multiple-alts/ma.conf

@@ -1,5 +1,5 @@
 [[annotation]]
 file="tests/multiple-alts/ma-db.vcf.gz"
-fields=["AF", "STR"]
-names= ["AF_out", "STR_out"]
-ops=["self", "self"]
+fields=["AF", "STR", "AF", "STR"]
+names= ["AF_out", "STR_out", "AF_by", "STR_by"]
+ops=["self", "self", "by_alt", "by_alt"]

vcfanno.go

@@ -25,7 +25,7 @@ import (
 	"github.com/brentp/xopen"
 )
 
-var VERSION = "0.2.7-beta2"
+var VERSION = "0.2.7-beta3"
 
 func envGet(name string, vdefault int) int {
 	sval := os.Getenv(name)