Skip to content

xa2multi.pl: reverse sequence and phred if orientations differ between primary and secondary alignment#5

Merged
lh3 merged 1 commit into
lh3:masterfrom
tobiasmarschall:master
Sep 7, 2011
Merged

xa2multi.pl: reverse sequence and phred if orientations differ between primary and secondary alignment#5
lh3 merged 1 commit into
lh3:masterfrom
tobiasmarschall:master

Conversation

@tobiasmarschall

Copy link
Copy Markdown

When a secondary alignment given in an XA tag differs in orientation from the primary one, the xa2multi.pl now gives the reverse complementary sequence and the reverse phred string.

… alignment has different orientation than primary alignment
lh3 pushed a commit that referenced this pull request Sep 7, 2011
xa2multi.pl: reverse sequence and phred if orientations differ between primary and secondary alignment
@lh3 lh3 merged commit beedc98 into lh3:master Sep 7, 2011
teepean added a commit to teepean/bwa that referenced this pull request Jun 1, 2026
lh3#5 real)

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
teepean added a commit to teepean/bwa that referenced this pull request Jun 1, 2026
…, bit-exact

Restructure bwa-aln-gpu into a producer/consumer pipeline: the main thread
owns the GPU (read -> MT preprocess -> upload -> kernel -> download has_hit,
serial on one stream so the shared per-warp global backing is never raced);
each chunk is handed to ONE in-order finisher thread doing the CPU reconcile +
output (.sai or fused samse), concurrent with the next chunk's GPU work. A
single ordered consumer preserves drand48 (repeat-hit) and output order ->
bit-exact. Structured multi-GPU-ready (replicate the GPU stage per device +
round-robin chunks; keep one ordered finisher) for when a 2nd card is added.

Full file (3.95M reads, fused alnse SAM): 174.2 s -> 160.9 s = 24,541 reads/s
(overlap hid ~13s of the ~18s CPU tail), records BIT-EXACT (md5 e2a6e1c9).
Both modes re-verified bit-exact under the pipeline (.sai eecf35c1, SAM
records identical). vs CPU bwa aln -t16 + samse = 895 s -> ~5.56x end-to-end.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants