123bc1d91d
put option operations in another file
2018-01-26 08:38:37 -05:00
dd18307e66
code backup
2018-01-25 21:52:49 -05:00
af1a871270
r658: gives a warning if -N0 is used
2018-01-19 08:33:20 -05:00
2b71181a37
r657: check -p ( #96 )
...
Well, in principle, every option should be checked. Will do when someone raise
issues...
2018-01-19 01:03:38 -05:00
33f8157961
r655: options to map to one strand of the ref #91
2018-01-16 10:34:30 -05:00
dc9e3dcf4a
r639: changed -O/-E validation
2017-12-30 20:39:29 -05:00
cc75c12905
r638: disabled scoring checking
...
I haven't figured out the exact bounds...
2017-12-30 07:50:40 -05:00
ab345e600b
r626: function to check incorrect scoring system
2017-12-13 12:23:43 -05:00
43960a8ca7
r621: --print-qname also shows kalloc status
2017-12-11 12:30:08 -05:00
f6608fe99c
r620: revamped thread-local memory management
...
* Don't preallocate sdust_buf or minizer list. kalloc should be fast enough -
benchmarks needed to confirm.
* Fixed a memory leak caused by divergence estimate (post v2.5)
* Reset the kalloc buffer after mapping a long query. This reduces peak memory
when large chunks of memory are allocated, at the cost of performance, though.
2017-12-11 12:11:10 -05:00
98a6e52c06
r618: heuristics to avoid tiny terminal exons
2017-12-11 00:57:55 -05:00
824712a4ee
r617: removed some unused code
2017-12-10 17:54:50 -05:00
98a999fe44
r611: added pseudocount when est divergence
2017-12-08 12:57:57 -05:00
2f693e8ca4
r609: bugfix - SDUST masking not working
2017-12-07 11:45:38 -05:00
704ff9f4c6
r607: estimate sequence divergence
...
Currently using the simplest method. There may be a more accurate estimate.
2017-12-06 16:14:39 -05:00
2f463b1db0
r573: prepare to generalize index
2017-11-11 19:54:06 -05:00
cd24dc8834
r545: removed option -i, not working well
2017-10-31 22:23:27 -04:00
fb8a1b5536
r542: tuning mapQ calculation
2017-10-31 14:25:09 -04:00
192217a10c
r539: use --splice-flank=yes by default
...
In human/mouse, the GTr..yAG pattern occurs to 91/92% of all GT-AG introns.
Modeling r..y clearly leads to higher accuracy. However, in SIRV, this
percentage is reduced to ~60%. The default "--splice --splice-flank=yes"
leads to lower accuracy. If someone benchmark minimap2 on SIRV, this would be
bad, but minimap2 is developed for practical applications, not for benchmarks.
I will live with that.
2017-10-28 22:29:55 -04:00
79b0caca95
r537: model the next base to GT/AG
...
[PMID:18688272] shows that the base following GT tends to be A or G (i.e. R) in
both human and yeast, and that the base preceeding AG tends to be C or T (i.e.
Y). In the new model, we pay no cost to GTr..yAG, but we pay half of the cost
if there is no r or y. This improves the junction accuracy when mapping to
human and mouse and decreases the accuacy when mapping to SIRV. My guess is
that SIRV does not honor this trend. Need to investigate in future.
Also in this commit, --cost-non-gt-ag is aliased to -C. The default is changed
to 9 instead of 5. I also added --splice-flank to enable the above model. This
may become the default once I confirm my hypothesis on SIRV.
2017-10-28 00:25:01 -04:00
d4b5dfc297
r533: added --no-pairing
...
to prevent the use of any pairing information for paired-end reads.
2017-10-23 14:09:32 -04:00
306e4541f8
Released minimap2-2.3 (r531)
2017-10-22 23:13:35 -04:00
bd04372873
r524: reverted to bwa-mem end bonus
...
and reduced the cost of clipping when filtering by identity
2017-10-20 16:57:31 -04:00
04cf4ebf5e
r518: increased the default -K to 500M
...
This helps multi-thread performance for ultra-long reads.
2017-10-17 13:21:29 -04:00
e6f525edaf
r512: option to filter poorly aligned reads
2017-10-16 10:38:22 -04:00
858213d513
r511: fixed wrong primary sam record
2017-10-12 23:02:18 -04:00
7c555f9b7e
r508: use two I/O threads for mapping
...
-x sr applies this option by default
2017-10-12 14:56:01 -04:00
7345621759
r499: end bonus working; DP region needs improve!
2017-10-11 00:14:25 -04:00
9396d9e11b
r452: typo in the last commit
2017-10-09 10:05:32 -04:00
198849a716
r491: an ambiguous base costs the same as gap ext
2017-10-09 09:59:42 -04:00
9fea4d16b3
r490: improved short-read extension heuristic
...
Now we find the best scoring ungapped seeded segment and then extend from it.
There is no gap filling for short reads.
2017-10-08 21:36:34 -04:00
61e56c941d
r488: parameter to control max fragment length
2017-10-07 23:54:32 -04:00
c6384ed2c8
r482: increased short-read bandwidth to 100
...
This has very minor effect on speed.
2017-10-06 10:20:32 -04:00
e0baf1ad54
r479: a bit code cleanup
2017-10-05 16:15:14 -04:00
f266092699
r478: simplied useless code, a tiny bit
2017-10-05 15:56:00 -04:00
9c5767f9ed
r477: renamed multi_seg to frag_mode
2017-10-05 15:48:17 -04:00
ae2adf04d4
r476: multi-file fragment mode working
2017-10-05 15:39:26 -04:00
5ab99eb26e
more accurate SAM flag
2017-10-05 10:59:38 -04:00
7cc4f6f965
r469: first step towards PE SAM
2017-10-05 10:38:09 -04:00
7d50e646dd
r466: detect multi-part index more smartly
...
though it might not work in an extremely rare case: the end of a sequence ends
at X*16384 and it is the last sequence in a batch. This can be resolved by
never letting the kstream_t buffer empty.
2017-10-04 17:32:58 -04:00
1554149158
r465: apply option -x before other options
2017-10-04 13:52:28 -04:00
2581c44a21
r463: optionally disable secondary hits
2017-10-04 13:24:41 -04:00
2a1e738a94
r461: randomize repetitive hits
2017-10-04 13:05:18 -04:00
cf55c84056
r460: added option --no-long-join
2017-10-04 12:08:44 -04:00
ee9b2773a8
r456: min chain score should >k-mer length
...
or chain_dp() wastes time on unnecessarily sorting chains with one k-mer.
2017-09-29 22:33:55 -04:00
04fb2c2ec0
r454: rechain with higher max_occ if no good chain
2017-09-29 19:24:32 -04:00
0d4ecd19ee
r453: avoid duplicated strcmp() for ava
2017-09-28 15:52:05 -04:00
9541052564
r447: paired-end mapping quality
...
not as good as I would hope...
2017-09-27 15:39:25 -04:00
7e0d70bfd3
r445: pair coordinate adjustment working
...
Next: mapq adjustment, which will be tricky...
2017-09-27 15:38:18 -04:00
a349d85280
r444: changed the way orientation is specified
...
The old model doesn't work with RF or RR orientation. The new model only works
with paired-end reads. For >2 segments, only FF is supported.
2017-09-27 12:33:10 -04:00
Heng Li