From 4b1fa532170fb94ea03a93102e346677e4dc819f Mon Sep 17 00:00:00 2001
From: Tom Clegg <tom@tomclegg.ca>
Date: Tue, 10 Aug 2021 15:26:37 -0400
Subject: [PATCH] Add filter options to export cmd.

refs #17562

Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
---
 export.go      |  6 ++++++
 export_test.go |  1 +
 exportnumpy.go |  4 +---
 filter.go      | 11 ++++++++---
 4 files changed, 16 insertions(+), 6 deletions(-)

diff --git a/export.go b/export.go
index 6bc1007b23..9546926c01 100644
--- a/export.go
+++ b/export.go
@@ -60,6 +60,7 @@ type exporter struct {
 	outputPerChrom bool
 	compress       bool
 	maxTileSize    int
+	filter         filter
 }
 
 func (cmd *exporter) RunCommand(prog string, args []string, stdin io.Reader, stdout, stderr io.Writer) int {
@@ -85,6 +86,7 @@ func (cmd *exporter) RunCommand(prog string, args []string, stdin io.Reader, std
 	flags.BoolVar(&cmd.compress, "z", false, "write gzip-compressed output files")
 	labelsFilename := flags.String("output-labels", "", "also output genome labels csv `file`")
 	flags.IntVar(&cmd.maxTileSize, "max-tile-size", 50000, "don't try to make annotations for tiles bigger than given `size`")
+	cmd.filter.Flags(flags)
 	err = flags.Parse(args)
 	if err == flag.ErrHelp {
 		err = nil
@@ -151,6 +153,7 @@ func (cmd *exporter) RunCommand(prog string, args []string, stdin io.Reader, std
 			"-output-dir", "/mnt/output",
 			"-z=" + fmt.Sprintf("%v", cmd.compress),
 		}
+		runner.Args = append(runner.Args, cmd.filter.Args()...)
 		var output string
 		output, err = runner.Run()
 		if err != nil {
@@ -182,6 +185,9 @@ func (cmd *exporter) RunCommand(prog string, args []string, stdin io.Reader, std
 		return 1
 	}
 
+	log.Infof("filtering: %+v", cmd.filter)
+	cmd.filter.Apply(tilelib)
+
 	names := cgnames(tilelib)
 	for _, name := range names {
 		cgs = append(cgs, CompactGenome{Name: name, Variants: tilelib.compactGenomes[name]})
diff --git a/export_test.go b/export_test.go
index dd60b52140..2cc0fbf366 100644
--- a/export_test.go
+++ b/export_test.go
@@ -132,6 +132,7 @@ chr2	469	.	GTGG	G	.	.	AC=1
 chr2	471	.	GG	AA	.	.	AC=1
 `))
 
+	c.Logf("export hgvs-numpy")
 	outdir := c.MkDir()
 	exited = (&exporter{}).RunCommand("export", []string{
 		"-local=true",
diff --git a/exportnumpy.go b/exportnumpy.go
index 00165b3db7..c8e9f7cfc5 100644
--- a/exportnumpy.go
+++ b/exportnumpy.go
@@ -95,11 +95,9 @@ func (cmd *exportNumpy) RunCommand(prog string, args []string, stdin io.Reader,
 			"-output-labels", "/mnt/output/labels.csv",
 			"-regions", *regionsFilename,
 			"-expand-regions", fmt.Sprintf("%d", *expandRegions),
-			"-max-variants", fmt.Sprintf("%d", cmd.filter.MaxVariants),
-			"-min-coverage", fmt.Sprintf("%f", cmd.filter.MinCoverage),
-			"-max-tag", fmt.Sprintf("%d", cmd.filter.MaxTag),
 			"-chunks", fmt.Sprintf("%d", *chunks),
 		}
+		runner.Args = append(runner.Args, cmd.filter.Args()...)
 		var output string
 		output, err = runner.Run()
 		if err != nil {
diff --git a/filter.go b/filter.go
index c66ea0ad0f..9d43abea60 100644
--- a/filter.go
+++ b/filter.go
@@ -65,9 +65,12 @@ func (f *filter) Apply(tilelib *tileLibrary) {
 	// f.MinCoverage.
 	mincov := int(2*f.MinCoverage*float64(len(tilelib.compactGenomes)) + 1)
 TAG:
-	for tag := 0; tag < len(tilelib.variant) && tag < f.MaxTag; tag++ {
+	for tag := 0; tag < len(tilelib.variant) && (tag < f.MaxTag || f.MaxTag < 0); tag++ {
 		tagcov := 0
 		for _, cg := range tilelib.compactGenomes {
+			if len(cg) < tag*2+2 {
+				continue
+			}
 			if cg[tag*2] > 0 {
 				tagcov++
 			}
@@ -79,8 +82,10 @@ TAG:
 			}
 		}
 		for _, cg := range tilelib.compactGenomes {
-			cg[tag*2] = 0
-			cg[tag*2+1] = 0
+			if len(cg) > tag*2 {
+				cg[tag*2] = 0
+				cg[tag*2+1] = 0
+			}
 		}
 	}
 
-- 
2.30.2