"flag"
"fmt"
"io"
+ "io/ioutil"
"net/http"
_ "net/http/pprof"
"os"
Name: "lightning slice-numpy",
Client: arvados.NewClientFromEnv(),
ProjectUUID: *projectUUID,
- RAM: 650000000000,
+ RAM: 750000000000,
VCPUs: 96,
Priority: *priority,
KeepCache: 2,
go func() {
defer throttleCPU.Release()
count := make(map[[blake2b.Size256]byte]int, len(variants))
+
+ rt := reftile[tag]
+ if rt != nil {
+ count[blake2b.Sum256(rt.tiledata)] = 0
+ }
+
for _, cg := range cgs {
idx := int(tag-tagstart) * 2
if idx < len(cg.Variants) {
remap[i] = rank[tv.Blake2b]
}
variantRemap[tag-tagstart] = remap
+ if rt != nil {
+ rt.variant = rank[blake2b.Sum256(rt.tiledata)]
+ }
}()
}
throttleCPU.Wait()
// mention it in annotations?)
continue
}
- variants, ok := seq[tag]
- if !ok {
- outcol++
- continue
- }
+ fmt.Fprintf(annow, "%d,%d,%d,=,%s,%d,,,\n", tag, outcol, rt.variant, rt.seqname, rt.pos)
+ variants := seq[tag]
reftilestr := strings.ToUpper(string(rt.tiledata))
remap := variantRemap[tag-tagstart]
done := make([]bool, len(variants))
for v, tv := range variants {
v := remap[v]
- if done[v] {
+ if v == rt.variant || done[v] {
continue
} else {
done[v] = true
}
if len(tv.Sequence) < taglen || !bytes.HasSuffix(rt.tiledata, tv.Sequence[len(tv.Sequence)-taglen:]) {
+ fmt.Fprintf(annow, "%d,%d,%d,,%s,%d,,,\n", tag, outcol, v, rt.seqname, rt.pos)
continue
}
if lendiff := len(rt.tiledata) - len(tv.Sequence); lendiff < -1000 || lendiff > 1000 {
+ fmt.Fprintf(annow, "%d,%d,%d,,%s,%d,,,\n", tag, outcol, v, rt.seqname, rt.pos)
continue
}
diffs, _ := hgvs.Diff(reftilestr, strings.ToUpper(string(tv.Sequence)), 0)
cols += len(chunk) / rows
}
out := make([]int16, rows*cols)
+ hgvsCols := map[string][2][]int16{} // hgvs -> [[g0,g1,g2,...], [g0,g1,g2,...]] (slice of genomes for each phase)
startcol := 0
for outIdx, chunk := range toMerge {
chunkcols := len(chunk) / rows
if len(line) == 0 {
continue
}
- fields := bytes.SplitN(line, []byte{','}, 3)
+ fields := bytes.SplitN(line, []byte{','}, 9)
+ tag, _ := strconv.Atoi(string(fields[0]))
incol, _ := strconv.Atoi(string(fields[1]))
- fmt.Fprintf(annow, "%s,%d,%s\n", fields[0], incol+startcol/2, fields[2])
+ tileVariant, _ := strconv.Atoi(string(fields[2]))
+ hgvsID := string(fields[3])
+ seqname := string(fields[4])
+ pos, _ := strconv.Atoi(string(fields[5]))
+ refseq := fields[6]
+ if hgvsID == "" {
+ // Null entry for un-diffable
+ // tile variant
+ continue
+ }
+ if hgvsID == "=" {
+ // Null entry for ref tile
+ continue
+ }
+ if mask != nil && !mask.Check(strings.TrimPrefix(seqname, "chr"), pos, pos+len(refseq)) {
+ // The tile intersects one of
+ // the selected regions, but
+ // this particular HGVS
+ // variant does not.
+ continue
+ }
+ hgvsColPair := hgvsCols[hgvsID]
+ if hgvsColPair[0] == nil {
+ // values in new columns start
+ // out as -1 ("no data yet")
+ // or 0 ("=ref") here, may
+ // change to 1 ("hgvs variant
+ // present") below, either on
+ // this line or a future line.
+ hgvsColPair = [2][]int16{make([]int16, len(cgnames)), make([]int16, len(cgnames))}
+ rt, ok := reftile[tagID(tag)]
+ if !ok {
+ err = fmt.Errorf("bug: seeing annotations for tag %d, but it has no reftile entry", tag)
+ return 1
+ }
+ for ph := 0; ph < 2; ph++ {
+ for row := 0; row < rows; row++ {
+ v := chunk[row*chunkcols+incol*2+ph]
+ if tileVariantID(v) == rt.variant {
+ hgvsColPair[ph][row] = 0
+ } else {
+ hgvsColPair[ph][row] = -1
+ }
+ }
+ }
+ hgvsCols[hgvsID] = hgvsColPair
+ hgvsref := hgvs.Variant{
+ Position: pos,
+ Ref: string(refseq),
+ New: string(refseq),
+ }
+ fmt.Fprintf(annow, "%d,%d,%d,%s:g.%s,%s,%d,%s,%s,%s\n", tag, incol+startcol/2, rt.variant, seqname, hgvsref.String(), seqname, pos, refseq, refseq, fields[8])
+ }
+ fmt.Fprintf(annow, "%d,%d,%d,%s,%s,%d,%s,%s,%s\n", tag, incol+startcol/2, tileVariant, hgvsID, seqname, pos, refseq, fields[7], fields[8])
+ for ph := 0; ph < 2; ph++ {
+ for row := 0; row < rows; row++ {
+ v := chunk[row*chunkcols+incol*2+ph]
+ if int(v) == tileVariant {
+ hgvsColPair[ph][row] = 1
+ }
+ }
+ }
}
startcol += chunkcols
if err != nil {
return 1
}
+ out = nil
+
+ cols = len(hgvsCols) * 2
+ log.Printf("building hgvs-based matrix: %d rows x %d cols", rows, cols)
+ out = make([]int16, rows*cols)
+ hgvsIDs := make([]string, 0, len(hgvsCols))
+ for hgvsID := range hgvsCols {
+ hgvsIDs = append(hgvsIDs, hgvsID)
+ }
+ sort.Strings(hgvsIDs)
+ var hgvsLabels bytes.Buffer
+ for idx, hgvsID := range hgvsIDs {
+ fmt.Fprintf(&hgvsLabels, "%d,%s\n", idx, hgvsID)
+ for ph := 0; ph < 2; ph++ {
+ hgvscol := hgvsCols[hgvsID][ph]
+ for row, val := range hgvscol {
+ out[row*cols+idx*2+ph] = val
+ }
+ }
+ }
+ err = writeNumpyInt16(fmt.Sprintf("%s/hgvs.npy", *outputDir), out, rows, cols)
+ if err != nil {
+ return 1
+ }
+
+ fnm := fmt.Sprintf("%s/hgvs.annotations.csv", *outputDir)
+ log.Printf("writing hgvs labels: %s", fnm)
+ err = ioutil.WriteFile(fnm, hgvsLabels.Bytes(), 0777)
+ if err != nil {
+ return 1
+ }
}
return 0
}
projects / lightning.git / blobdiff