ref2genome.go

   1 package main
   2
   3 import (
   4         "bufio"
   5         "bytes"
   6         "compress/gzip"
   7         "errors"
   8         "flag"
   9         "fmt"
  10         "io"
  11         "net/http"
  12         _ "net/http/pprof"
  13         "os"
  14         "strings"
  15
  16         "git.arvados.org/arvados.git/sdk/go/arvados"
  17         log "github.com/sirupsen/logrus"
  18 )
  19
  20 type ref2genome struct {
  21         refFile        string
  22         projectUUID    string
  23         outputFilename string
  24         runLocal       bool
  25 }
  26
  27 func (cmd *ref2genome) RunCommand(prog string, args []string, stdin io.Reader, stdout, stderr io.Writer) int {
  28         var err error
  29         defer func() {
  30                 if err != nil {
  31                         fmt.Fprintf(stderr, "%s\n", err)
  32                 }
  33         }()
  34         flags := flag.NewFlagSet("", flag.ContinueOnError)
  35         flags.SetOutput(stderr)
  36         flags.StringVar(&cmd.refFile, "ref", "", "reference fasta `file`")
  37         flags.StringVar(&cmd.projectUUID, "project", "", "project `UUID` for containers and output data")
  38         flags.StringVar(&cmd.outputFilename, "o", "", "output filename")
  39         flags.BoolVar(&cmd.runLocal, "local", false, "run on local host (default: run in an arvados container)")
  40         priority := flags.Int("priority", 500, "container request priority")
  41         pprof := flags.String("pprof", "", "serve Go profile data at http://`[addr]:port`")
  42         err = flags.Parse(args)
  43         if err == flag.ErrHelp {
  44                 err = nil
  45                 return 0
  46         } else if err != nil {
  47                 return 2
  48         } else if cmd.refFile == "" {
  49                 err = errors.New("reference data (-ref) not specified")
  50                 return 2
  51         }
  52
  53         if *pprof != "" {
  54                 go func() {
  55                         log.Println(http.ListenAndServe(*pprof, nil))
  56                 }()
  57         }
  58
  59         if !cmd.runLocal {
  60                 if cmd.outputFilename != "" {
  61                         err = errors.New("cannot specify output filename in non-local mode")
  62                         return 2
  63                 }
  64                 runner := arvadosContainerRunner{
  65                         Name:        "lightning ref2genome",
  66                         Client:      arvados.NewClientFromEnv(),
  67                         ProjectUUID: cmd.projectUUID,
  68                         RAM:         1 << 30,
  69                         Priority:    *priority,
  70                         VCPUs:       1,
  71                 }
  72                 err = runner.TranslatePaths(&cmd.refFile)
  73                 if err != nil {
  74                         return 1
  75                 }
  76                 runner.Args = []string{"ref2genome", "-local=true", "-ref", cmd.refFile, "-o", "/mnt/output/ref.genome"}
  77                 var output string
  78                 output, err = runner.Run()
  79                 if err != nil {
  80                         return 1
  81                 }
  82                 fmt.Fprintln(stdout, output+"/ref.genome")
  83                 return 0
  84         }
  85
  86         var out io.WriteCloser
  87         if cmd.outputFilename == "" {
  88                 out = nopCloser{stdout}
  89         } else {
  90                 out, err = os.OpenFile(cmd.outputFilename, os.O_CREATE|os.O_TRUNC|os.O_WRONLY, 0666)
  91                 if err != nil {
  92                         return 1
  93                 }
  94         }
  95         f, err := os.Open(cmd.refFile)
  96         if err != nil {
  97                 return 1
  98         }
  99         defer f.Close()
 100         var in io.Reader
 101         if strings.HasSuffix(cmd.refFile, ".gz") {
 102                 in, err = gzip.NewReader(f)
 103                 if err != nil {
 104                         return 1
 105                 }
 106         } else {
 107                 in = f
 108         }
 109         label, seqlen := "", 0
 110         scanner := bufio.NewScanner(in)
 111         for scanner.Scan() {
 112                 buf := scanner.Bytes()
 113                 if len(buf) > 0 && buf[0] == '>' {
 114                         if label != "" {
 115                                 fmt.Fprintf(out, "%s\t%d\n", label, seqlen)
 116                         }
 117                         label = strings.TrimSpace(string(buf[1:]))
 118                         seqlen = 0
 119                 } else {
 120                         seqlen += len(bytes.TrimSpace(buf))
 121                 }
 122         }
 123         if label != "" {
 124                 fmt.Fprintf(out, "%s\t%d\n", label, seqlen)
 125         }
 126         if err = scanner.Err(); err != nil {
 127                 return 1
 128         }
 129         if err = out.Close(); err != nil {
 130                 return 1
 131         }
 132         return 0
 133 }