"flag"
"fmt"
"io"
- "log"
"net/http"
_ "net/http/pprof"
"os"
"sync"
"sync/atomic"
"time"
+
+ "git.arvados.org/arvados.git/sdk/go/arvados"
+ log "github.com/sirupsen/logrus"
)
type importer struct {
tagLibraryFile string
refFile string
+ outputFile string
+ projectUUID string
+ runLocal bool
+ skipOOO bool
encoder *gob.Encoder
}
flags.SetOutput(stderr)
flags.StringVar(&cmd.tagLibraryFile, "tag-library", "", "tag library fasta `file`")
flags.StringVar(&cmd.refFile, "ref", "", "reference fasta `file`")
+ flags.StringVar(&cmd.outputFile, "o", "-", "output `file`")
+ flags.StringVar(&cmd.projectUUID, "project", "", "project `UUID` for output data")
+ flags.BoolVar(&cmd.runLocal, "local", false, "run on local host (default: run in an arvados container)")
+ flags.BoolVar(&cmd.skipOOO, "skip-ooo", false, "skip out-of-order tags")
+ priority := flags.Int("priority", 500, "container request priority")
pprof := flags.String("pprof", "", "serve Go profile data at http://`[addr]:port`")
+ loglevel := flags.String("loglevel", "info", "logging threshold (trace, debug, info, warn, error, fatal, or panic)")
err = flags.Parse(args)
if err == flag.ErrHelp {
err = nil
}()
}
+ lvl, err := log.ParseLevel(*loglevel)
+ if err != nil {
+ return 2
+ }
+ log.SetLevel(lvl)
+
+ if !cmd.runLocal {
+ runner := arvadosContainerRunner{
+ Name: "lightning import",
+ Client: arvados.NewClientFromEnv(),
+ ProjectUUID: cmd.projectUUID,
+ RAM: 60000000000,
+ VCPUs: 16,
+ Priority: *priority,
+ }
+ err = runner.TranslatePaths(&cmd.tagLibraryFile, &cmd.refFile, &cmd.outputFile)
+ if err != nil {
+ return 1
+ }
+ inputs := flags.Args()
+ for i := range inputs {
+ err = runner.TranslatePaths(&inputs[i])
+ if err != nil {
+ return 1
+ }
+ }
+ if cmd.outputFile == "-" {
+ cmd.outputFile = "/mnt/output/library.gob"
+ } else {
+ // Not yet implemented, but this should write
+ // the collection to an existing collection,
+ // possibly even an in-place update.
+ err = errors.New("cannot specify output file in container mode: not implemented")
+ return 1
+ }
+ runner.Args = append([]string{"import", "-local=true", "-loglevel=" + *loglevel, fmt.Sprintf("-skip-ooo=%v", cmd.skipOOO), "-tag-library", cmd.tagLibraryFile, "-ref", cmd.refFile, "-o", cmd.outputFile}, inputs...)
+ var output string
+ output, err = runner.Run()
+ if err != nil {
+ return 1
+ }
+ fmt.Fprintln(stdout, output+"/library.gob")
+ return 0
+ }
+
infiles, err := listInputFiles(flags.Args())
if err != nil {
return 1
return 1
}
go func() {
- for range time.Tick(10 * time.Second) {
+ for range time.Tick(10 * time.Minute) {
log.Printf("tilelib.Len() == %d", tilelib.Len())
}
}()
- w := bufio.NewWriter(stdout)
- cmd.encoder = gob.NewEncoder(w)
+
+ var output io.WriteCloser
+ if cmd.outputFile == "-" {
+ output = nopCloser{stdout}
+ } else {
+ output, err = os.OpenFile(cmd.outputFile, os.O_CREATE|os.O_WRONLY, 0777)
+ if err != nil {
+ return 1
+ }
+ defer output.Close()
+ }
+ bufw := bufio.NewWriter(output)
+ cmd.encoder = gob.NewEncoder(bufw)
+
err = cmd.tileInputs(tilelib, infiles)
if err != nil {
return 1
}
- err = w.Flush()
+ err = bufw.Flush()
+ if err != nil {
+ return 1
+ }
+ err = output.Close()
if err != nil {
return 1
}
return nil, fmt.Errorf("cannot tile: tag library is empty")
}
log.Printf("tag library %s load done", cmd.tagLibraryFile)
- return &tileLibrary{taglib: &taglib}, nil
+ return &tileLibrary{taglib: &taglib, skipOOO: cmd.skipOOO}, nil
}
func listInputFiles(paths []string) (files []string, err error) {
log.Printf("%s starting", infile)
defer log.Printf("%s done", infile)
tseqs, err := cmd.tileFasta(tilelib, infile)
- variants[0] = tseqs.Variants()
+ var kept, dropped int
+ variants[0], kept, dropped = tseqs.Variants()
+ log.Printf("%s found %d unique tags plus %d repeats", infile, kept, dropped)
return err
}
infile2 := regexp.MustCompile(`\.1\.fasta(\.gz)?$`).ReplaceAllString(infile, `.2.fasta$1`)
log.Printf("%s starting", infile2)
defer log.Printf("%s done", infile2)
tseqs, err := cmd.tileFasta(tilelib, infile2)
- variants[1] = tseqs.Variants()
+ var kept, dropped int
+ variants[1], kept, dropped = tseqs.Variants()
+ log.Printf("%s found %d unique tags plus %d repeats", infile, kept, dropped)
return err
}
} else {
log.Printf("%s phase %d starting", infile, phase+1)
defer log.Printf("%s phase %d done", infile, phase+1)
tseqs, err := cmd.tileGVCF(tilelib, infile, phase)
- variants[phase] = tseqs.Variants()
+ var kept, dropped int
+ variants[phase], kept, dropped = tseqs.Variants()
+ log.Printf("%s phase %d found %d unique tags plus %d repeats", infile, phase+1, kept, dropped)
return err
}
}
}
flat := make([]tileVariantID, ntags*2)
for i := 0; i < ntags; i++ {
- flat[i*2] = variants[0][i]
- flat[i*2+1] = variants[1][i]
+ for hap := 0; hap < 2; hap++ {
+ if i < len(variants[hap]) {
+ flat[i*2+hap] = variants[hap][i]
+ }
+ }
}
err := cmd.encoder.Encode(LibraryEntry{
CompactGenomes: []CompactGenome{{Name: infile, Variants: flat}},
}
go close(todo)
var tileJobs sync.WaitGroup
- running := int64(runtime.NumCPU())
- for i := 0; i < runtime.NumCPU(); i++ {
+ var running int64
+ for i := 0; i < runtime.NumCPU()*9/8+1; i++ {
tileJobs.Add(1)
+ atomic.AddInt64(&running, 1)
go func() {
defer tileJobs.Done()
defer atomic.AddInt64(&running, -1)
projects / lightning.git / blobdiff