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         "net/http"
  14         _ "net/http/pprof"
  15         "os"
  16         "runtime"
  17         "sort"
  18         "strconv"
  19         "strings"
  20         "sync"
  21
  22         "git.arvados.org/arvados.git/sdk/go/arvados"
  23         log "github.com/sirupsen/logrus"
  24 )
  25
  26 type anno2vcf struct {
  27 }
  28
  29 func (cmd *anno2vcf) RunCommand(prog string, args []string, stdin io.Reader, stdout, stderr io.Writer) int {
  30         var err error
  31         defer func() {
  32                 if err != nil {
  33                         fmt.Fprintf(stderr, "%s\n", err)
  34                 }
  35         }()
  36         flags := flag.NewFlagSet("", flag.ContinueOnError)
  37         flags.SetOutput(stderr)
  38         pprof := flags.String("pprof", "", "serve Go profile data at http://`[addr]:port`")
  39         runlocal := flags.Bool("local", false, "run on local host (default: run in an arvados container)")
  40         projectUUID := flags.String("project", "", "project `UUID` for output data")
  41         priority := flags.Int("priority", 500, "container request priority")
  42         inputDir := flags.String("input-dir", "./in", "input `directory`")
  43         outputDir := flags.String("output-dir", "./out", "output `directory`")
  44         err = flags.Parse(args)
  45         if err == flag.ErrHelp {
  46                 err = nil
  47                 return 0
  48         } else if err != nil {
  49                 return 2
  50         }
  51
  52         if *pprof != "" {
  53                 go func() {
  54                         log.Println(http.ListenAndServe(*pprof, nil))
  55                 }()
  56         }
  57
  58         if !*runlocal {
  59                 runner := arvadosContainerRunner{
  60                         Name:        "lightning anno2vcf",
  61                         Client:      arvados.NewClientFromEnv(),
  62                         ProjectUUID: *projectUUID,
  63                         RAM:         500000000000,
  64                         VCPUs:       64,
  65                         Priority:    *priority,
  66                         KeepCache:   2,
  67                         APIAccess:   true,
  68                 }
  69                 err = runner.TranslatePaths(inputDir)
  70                 if err != nil {
  71                         return 1
  72                 }
  73                 runner.Args = []string{"anno2vcf", "-local=true",
  74                         "-pprof", ":6060",
  75                         "-input-dir", *inputDir,
  76                         "-output-dir", "/mnt/output",
  77                 }
  78                 var output string
  79                 output, err = runner.Run()
  80                 if err != nil {
  81                         return 1
  82                 }
  83                 fmt.Fprintln(stdout, output)
  84                 return 0
  85         }
  86
  87         d, err := open(*inputDir)
  88         if err != nil {
  89                 log.Print(err)
  90                 return 1
  91         }
  92         defer d.Close()
  93         fis, err := d.Readdir(-1)
  94         if err != nil {
  95                 log.Print(err)
  96                 return 1
  97         }
  98         d.Close()
  99         sort.Slice(fis, func(i, j int) bool { return fis[i].Name() < fis[j].Name() })
 100
 101         type call struct {
 102                 tile      int
 103                 variant   int
 104                 position  int
 105                 deletion  []byte
 106                 insertion []byte
 107                 hgvsID    []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                         f, err := open(filename)
 121                         if err != nil {
 122                                 return err
 123                         }
 124                         defer f.Close()
 125                         buf, err := io.ReadAll(f)
 126                         if err != nil {
 127                                 return fmt.Errorf("%s: %s", filename, err)
 128                         }
 129                         f.Close()
 130                         lines := bytes.Split(buf, []byte{'\n'})
 131                         calls := map[string][]*call{}
 132                         for lineIdx, line := range lines {
 133                                 if len(line) == 0 {
 134                                         continue
 135                                 }
 136                                 if lineIdx&0xff == 0 && thr.Err() != nil {
 137                                         return nil
 138                                 }
 139                                 fields := bytes.Split(line, []byte{','})
 140                                 if len(fields) < 8 {
 141                                         return fmt.Errorf("%s line %d: wrong number of fields (%d < %d): %q", fi.Name(), lineIdx+1, len(fields), 8, line)
 142                                 }
 143                                 hgvsID := fields[3]
 144                                 if len(hgvsID) < 2 {
 145                                         // "=" reference or ""
 146                                         // non-diffable tile variant
 147                                         continue
 148                                 }
 149                                 tile, _ := strconv.ParseInt(string(fields[0]), 10, 64)
 150                                 variant, _ := strconv.ParseInt(string(fields[2]), 10, 64)
 151                                 position, _ := strconv.ParseInt(string(fields[5]), 10, 64)
 152                                 seq := string(fields[4])
 153                                 if calls[seq] == nil {
 154                                         calls[seq] = make([]*call, 0, len(lines)/50)
 155                                 }
 156                                 del := fields[6]
 157                                 ins := fields[7]
 158                                 if (len(del) == 0 || len(ins) == 0) && len(fields) >= 9 {
 159                                         // "123,,AA,T" means 123insAA
 160                                         // preceded by T. We record it
 161                                         // here as "122 T TAA" to
 162                                         // avoid writing an empty
 163                                         // "ref" field in our
 164                                         // VCF. Similarly, we record
 165                                         // deletions as "122 TAA T"
 166                                         // rather than "123 AA .".
 167                                         del = append(append(make([]byte, 0, len(fields[8])+len(del)), fields[8]...), del...)
 168                                         ins = append(append(make([]byte, 0, len(fields[8])+len(ins)), fields[8]...), ins...)
 169                                         position -= int64(len(fields[8]))
 170                                 } else {
 171                                         del = append([]byte(nil), del...)
 172                                         ins = append([]byte(nil), ins...)
 173                                 }
 174                                 calls[seq] = append(calls[seq], &call{
 175                                         tile:      int(tile),
 176                                         variant:   int(variant),
 177                                         position:  int(position),
 178                                         deletion:  del,
 179                                         insertion: ins,
 180                                         hgvsID:    hgvsID,
 181                                 })
 182                         }
 183                         mtx.Lock()
 184                         for seq, seqcalls := range calls {
 185                                 allcalls[seq] = append(allcalls[seq], seqcalls...)
 186                         }
 187                         mtx.Unlock()
 188                         return nil
 189                 })
 190         }
 191         err = thr.Wait()
 192         if err != nil {
 193                 return 1
 194         }
 195         thr = throttle{Max: len(allcalls)}
 196         for seq, seqcalls := range allcalls {
 197                 seq, seqcalls := seq, seqcalls
 198                 thr.Go(func() error {
 199                         log.Printf("%s: sorting", seq)
 200                         sort.Slice(seqcalls, func(i, j int) bool {
 201                                 ii, jj := seqcalls[i], seqcalls[j]
 202                                 if cmp := ii.position - jj.position; cmp != 0 {
 203                                         return cmp < 0
 204                                 }
 205                                 if cmp := len(ii.deletion) - len(jj.deletion); cmp != 0 {
 206                                         return cmp < 0
 207                                 }
 208                                 if cmp := bytes.Compare(ii.insertion, jj.insertion); cmp != 0 {
 209                                         return cmp < 0
 210                                 }
 211                                 if cmp := ii.tile - jj.tile; cmp != 0 {
 212                                         return cmp < 0
 213                                 }
 214                                 return ii.variant < jj.variant
 215                         })
 216
 217                         vcfFilename := fmt.Sprintf("%s/annotations.%s.vcf", *outputDir, seq)
 218                         log.Printf("%s: writing %s", seq, vcfFilename)
 219
 220                         f, err := os.Create(vcfFilename)
 221                         if err != nil {
 222                                 return err
 223                         }
 224                         defer f.Close()
 225                         bufw := bufio.NewWriterSize(f, 1<<20)
 226                         _, err = fmt.Fprintf(bufw, `##fileformat=VCFv4.0
 227 ##INFO=<ID=TV,Number=.,Type=String,Description="tile-variant">
 228 #CHROM  POS     ID      REF     ALT     QUAL    FILTER  INFO
 229 `)
 230                         if err != nil {
 231                                 return err
 232                         }
 233                         placeholder := []byte{'.'}
 234                         for i := 0; i < len(seqcalls); {
 235                                 call := seqcalls[i]
 236                                 i++
 237                                 info := fmt.Sprintf("TV=,%d-%d,", call.tile, call.variant)
 238                                 for i < len(seqcalls) &&
 239                                         call.position == seqcalls[i].position &&
 240                                         len(call.deletion) == len(seqcalls[i].deletion) &&
 241                                         bytes.Equal(call.insertion, seqcalls[i].insertion) {
 242                                         call = seqcalls[i]
 243                                         i++
 244                                         info += fmt.Sprintf("%d-%d,", call.tile, call.variant)
 245                                 }
 246                                 deletion := call.deletion
 247                                 if len(deletion) == 0 {
 248                                         deletion = placeholder
 249                                 }
 250                                 insertion := call.insertion
 251                                 if len(insertion) == 0 {
 252                                         insertion = placeholder
 253                                 }
 254                                 _, err = fmt.Fprintf(bufw, "%s\t%d\t%s\t%s\t%s\t.\t.\t%s\n", seq, call.position, call.hgvsID, deletion, insertion, info)
 255                                 if err != nil {
 256                                         return err
 257                                 }
 258                         }
 259                         err = bufw.Flush()
 260                         if err != nil {
 261                                 return err
 262                         }
 263                         err = f.Close()
 264                         if err != nil {
 265                                 return err
 266                         }
 267                         log.Printf("%s: done", seq)
 268                         return nil
 269                 })
 270         }
 271         err = thr.Wait()
 272         if err != nil {
 273                 return 1
 274         }
 275         return 0
 276 }