"flag"
"fmt"
"io"
+ "io/ioutil"
"net/http"
_ "net/http/pprof"
"os"
return 1
}
if isdup[libref.Tag] {
- log.Printf("dropping reference tile %+v from %s, tag not unique", libref, seqname)
+ log.Printf("dropping reference tile %+v from %s @ %d, tag not unique", libref, seqname, pos)
} else if reftile[libref.Tag] != nil {
- log.Printf("dropping reference tile %+v from %s, tag not unique", tileLibRef{Tag: libref.Tag, Variant: reftile[libref.Tag].variant}, reftile[libref.Tag].seqname)
+ log.Printf("dropping reference tile %+v from %s @ %d, tag not unique", tileLibRef{Tag: libref.Tag, Variant: reftile[libref.Tag].variant}, reftile[libref.Tag].seqname, reftile[libref.Tag].pos)
delete(reftile, libref.Tag)
- log.Printf("dropping reference tile %+v from %s, tag not unique", libref, seqname)
+ log.Printf("dropping reference tile %+v from %s @ %d, tag not unique", libref, seqname, pos)
isdup[libref.Tag] = true
} else {
reftile[libref.Tag] = &reftileinfo{
}
variants, ok := seq[tag]
if !ok {
- outcol++
- continue
+ // how could we even have a reftile if there is no sequence data??
+ return fmt.Errorf("bug: have no variants for tag %d but reftile is %+v", tag, rt)
}
reftilestr := strings.ToUpper(string(rt.tiledata))
remap := variantRemap[tag-tagstart]
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)
+ hgvs := 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{','}, 8)
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 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
+ }
+ fmt.Fprintf(annow, "%s,%d,%d,%s,%s,%d,%s,%s\n", fields[0], incol+startcol/2, tileVariant, hgvsID, seqname, pos, refseq, fields[7])
+ hgvscols := hgvs[hgvsID]
+ if hgvscols[0] == nil {
+ hgvscols = [2][]int16{make([]int16, len(cgnames)), make([]int16, len(cgnames))}
+ hgvs[hgvsID] = hgvscols
+ }
+ for ph := 0; ph < 2; ph++ {
+ for row := 0; row < rows; row++ {
+ v := chunk[row*chunkcols+incol*2+ph]
+ if int(v) == tileVariant {
+ if len(hgvsID) == 0 {
+ // we have the tile variant sequence, but the diff against ref didn't work out (see lendiff above)
+ hgvscols[ph][row] = -2
+ } else {
+ hgvscols[ph][row] = 1
+ }
+ } else if v < 0 {
+ hgvscols[ph][row] = -1
+ }
+ }
+ }
}
startcol += chunkcols
if err != nil {
return 1
}
+ out = nil
+
+ cols = len(hgvs) * 2
+ log.Printf("building hgvs-based matrix: %d rows x %d cols", rows, cols)
+ out = make([]int16, rows*cols)
+ hgvsIDs := make([]string, 0, len(hgvs))
+ for hgvsID := range hgvs {
+ 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 := hgvs[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
}