anno2vcf.go

   1 // Copyright (C) The Lightning Authors. All rights reserved.
   2 //
   3 // SPDX-License-Identifier: AGPL-3.0
   4
   5 package lightning
   6
   7 import (
   8         "bufio"
   9         "bytes"
  10         "flag"
  11         "fmt"
  12         "io"
  13         "io/ioutil"
  14         "net/http"
  15         _ "net/http/pprof"
  16         "os"
  17         "runtime"
  18         "sort"
  19         "strconv"
  20         "strings"
  21         "sync"
  22
  23         "git.arvados.org/arvados.git/sdk/go/arvados"
  24         log "github.com/sirupsen/logrus"
  25 )
  26
  27 type anno2vcf struct {
  28 }
  29
  30 func (cmd *anno2vcf) RunCommand(prog string, args []string, stdin io.Reader, stdout, stderr io.Writer) int {
  31         var err error
  32         defer func() {
  33                 if err != nil {
  34                         fmt.Fprintf(stderr, "%s\n", err)
  35                 }
  36         }()
  37         flags := flag.NewFlagSet("", flag.ContinueOnError)
  38         flags.SetOutput(stderr)
  39         pprof := flags.String("pprof", "", "serve Go profile data at http://`[addr]:port`")
  40         runlocal := flags.Bool("local", false, "run on local host (default: run in an arvados container)")
  41         projectUUID := flags.String("project", "", "project `UUID` for output data")
  42         priority := flags.Int("priority", 500, "container request priority")
  43         inputDir := flags.String("input-dir", "./in", "input `directory`")
  44         outputDir := flags.String("output-dir", "./out", "output `directory`")
  45         err = flags.Parse(args)
  46         if err == flag.ErrHelp {
  47                 err = nil
  48                 return 0
  49         } else if err != nil {
  50                 return 2
  51         }
  52
  53         if *pprof != "" {
  54                 go func() {
  55                         log.Println(http.ListenAndServe(*pprof, nil))
  56                 }()
  57         }
  58
  59         if !*runlocal {
  60                 runner := arvadosContainerRunner{
  61                         Name:        "lightning anno2vcf",
  62                         Client:      arvados.NewClientFromEnv(),
  63                         ProjectUUID: *projectUUID,
  64                         RAM:         500000000000,
  65                         VCPUs:       64,
  66                         Priority:    *priority,
  67                         KeepCache:   2,
  68                         APIAccess:   true,
  69                 }
  70                 err = runner.TranslatePaths(inputDir)
  71                 if err != nil {
  72                         return 1
  73                 }
  74                 runner.Args = []string{"anno2vcf", "-local=true",
  75                         "-pprof", ":6060",
  76                         "-input-dir", *inputDir,
  77                         "-output-dir", "/mnt/output",
  78                 }
  79                 var output string
  80                 output, err = runner.Run()
  81                 if err != nil {
  82                         return 1
  83                 }
  84                 fmt.Fprintln(stdout, output)
  85                 return 0
  86         }
  87
  88         d, err := open(*inputDir)
  89         if err != nil {
  90                 log.Print(err)
  91                 return 1
  92         }
  93         defer d.Close()
  94         fis, err := d.Readdir(-1)
  95         if err != nil {
  96                 log.Print(err)
  97                 return 1
  98         }
  99         d.Close()
 100         sort.Slice(fis, func(i, j int) bool { return fis[i].Name() < fis[j].Name() })
 101
 102         type call struct {
 103                 tile      int
 104                 variant   int
 105                 position  int
 106                 deletion  []byte
 107                 insertion []byte
 108         }
 109         allcalls := map[string][]*call{}
 110         var mtx sync.Mutex
 111         thr := throttle{Max: runtime.GOMAXPROCS(0)}
 112         log.Print("reading input files")
 113         for _, fi := range fis {
 114                 if !strings.HasSuffix(fi.Name(), "annotations.csv") {
 115                         continue
 116                 }
 117                 filename := *inputDir + "/" + fi.Name()
 118                 thr.Go(func() error {
 119                         log.Printf("reading %s", filename)
 120                         buf, err := ioutil.ReadFile(filename)
 121                         if err != nil {
 122                                 return fmt.Errorf("%s: %s", filename, err)
 123                         }
 124                         lines := bytes.Split(buf, []byte{'\n'})
 125                         calls := map[string][]*call{}
 126                         for lineIdx, line := range lines {
 127                                 if len(line) == 0 {
 128                                         continue
 129                                 }
 130                                 if lineIdx & ^0xfff == 0 && thr.Err() != nil {
 131                                         return nil
 132                                 }
 133                                 fields := bytes.Split(line, []byte{','})
 134                                 if len(fields) < 8 {
 135                                         return fmt.Errorf("%s line %d: wrong number of fields (%d < %d): %q", fi.Name(), lineIdx+1, len(fields), 8, line)
 136                                 }
 137                                 tile, _ := strconv.ParseInt(string(fields[0]), 10, 64)
 138                                 variant, _ := strconv.ParseInt(string(fields[2]), 10, 64)
 139                                 position, _ := strconv.ParseInt(string(fields[5]), 10, 64)
 140                                 seq := string(fields[4])
 141                                 if calls[seq] == nil {
 142                                         calls[seq] = make([]*call, 0, len(lines)/50)
 143                                 }
 144                                 del := fields[6]
 145                                 ins := fields[7]
 146                                 if len(del) == 0 && len(fields) >= 9 {
 147                                         // "123,,AA,T" means 123insAA
 148                                         // preceded by T. We record it
 149                                         // here as 122TdelinsTAA to
 150                                         // avoid writing an empty
 151                                         // "ref" field in our VCF.
 152                                         del = append([]byte(nil), fields[8]...)
 153                                         ins = append(append([]byte(nil), del...), ins...)
 154                                         position -= int64(len(del))
 155                                 } else {
 156                                         del = append([]byte(nil), del...)
 157                                         ins = append([]byte(nil), ins...)
 158                                 }
 159                                 calls[seq] = append(calls[seq], &call{
 160                                         tile:      int(tile),
 161                                         variant:   int(variant),
 162                                         position:  int(position),
 163                                         deletion:  del,
 164                                         insertion: ins,
 165                                 })
 166                         }
 167                         mtx.Lock()
 168                         for seq, seqcalls := range calls {
 169                                 allcalls[seq] = append(allcalls[seq], seqcalls...)
 170                         }
 171                         mtx.Unlock()
 172                         return nil
 173                 })
 174         }
 175         err = thr.Wait()
 176         if err != nil {
 177                 return 1
 178         }
 179         thr = throttle{Max: len(allcalls)}
 180         for seq, seqcalls := range allcalls {
 181                 seq, seqcalls := seq, seqcalls
 182                 thr.Go(func() error {
 183                         log.Printf("%s: sorting", seq)
 184                         sort.Slice(seqcalls, func(i, j int) bool {
 185                                 ii, jj := seqcalls[i], seqcalls[j]
 186                                 if cmp := ii.position - jj.position; cmp != 0 {
 187                                         return cmp < 0
 188                                 }
 189                                 if cmp := len(ii.deletion) - len(jj.deletion); cmp != 0 {
 190                                         return cmp < 0
 191                                 }
 192                                 if cmp := bytes.Compare(ii.insertion, jj.insertion); cmp != 0 {
 193                                         return cmp < 0
 194                                 }
 195                                 if cmp := ii.tile - jj.tile; cmp != 0 {
 196                                         return cmp < 0
 197                                 }
 198                                 return ii.variant < jj.variant
 199                         })
 200
 201                         vcfFilename := fmt.Sprintf("%s/annotations.%s.vcf", *outputDir, seq)
 202                         log.Printf("%s: writing %s", seq, vcfFilename)
 203
 204                         f, err := os.Create(vcfFilename)
 205                         if err != nil {
 206                                 return err
 207                         }
 208                         defer f.Close()
 209                         bufw := bufio.NewWriterSize(f, 1<<20)
 210                         _, err = fmt.Fprintf(bufw, `##fileformat=VCFv4.0
 211 ##INFO=<ID=TV,Number=.,Type=String,Description="tile-variant">
 212 #CHROM  POS     ID      REF     ALT     QUAL    FILTER  INFO
 213 `)
 214                         if err != nil {
 215                                 return err
 216                         }
 217                         placeholder := []byte{'.'}
 218                         for i := 0; i < len(seqcalls); {
 219                                 call := seqcalls[i]
 220                                 i++
 221                                 info := fmt.Sprintf("TV=,%d-%d,", call.tile, call.variant)
 222                                 for i < len(seqcalls) &&
 223                                         call.position == seqcalls[i].position &&
 224                                         len(call.deletion) == len(seqcalls[i].deletion) &&
 225                                         bytes.Equal(call.insertion, seqcalls[i].insertion) {
 226                                         call = seqcalls[i]
 227                                         i++
 228                                         info += fmt.Sprintf("%d-%d,", call.tile, call.variant)
 229                                 }
 230                                 deletion := call.deletion
 231                                 if len(deletion) == 0 {
 232                                         deletion = placeholder
 233                                 }
 234                                 insertion := call.insertion
 235                                 if len(insertion) == 0 {
 236                                         insertion = placeholder
 237                                 }
 238                                 _, err = fmt.Fprintf(bufw, "%s\t%d\t.\t%s\t%s\t.\t.\t%s\n", seq, call.position, deletion, insertion, info)
 239                                 if err != nil {
 240                                         return err
 241                                 }
 242                         }
 243                         err = bufw.Flush()
 244                         if err != nil {
 245                                 return err
 246                         }
 247                         err = f.Close()
 248                         if err != nil {
 249                                 return err
 250                         }
 251                         log.Printf("%s: done", seq)
 252                         return nil
 253                 })
 254         }
 255         err = thr.Wait()
 256         if err != nil {
 257                 return 1
 258         }
 259         return 0
 260 }