]>
git.donarmstrong.com Git - samtools.git/log
Heng Li [Mon, 15 Nov 2010 05:30:18 +0000 (05:30 +0000)]
* samtools-0.1.9-16 (r822)
* keep the raw depth because in indel calling, DP4 may be way off the true depth
Heng Li [Sat, 13 Nov 2010 06:18:31 +0000 (06:18 +0000)]
* samtools-0.1.9-15 (r821)
* calmd: write BQ
* skip realignment if BQ is present
Heng Li [Sat, 13 Nov 2010 06:08:26 +0000 (06:08 +0000)]
* samtools-0.1.9-14 (r820)
* penalize reads with excessive differences in indel calling
Heng Li [Sat, 13 Nov 2010 02:36:27 +0000 (02:36 +0000)]
* samtools-0.1.9-13 (r819)
* fixed a bug in pileup given refskip
Heng Li [Fri, 12 Nov 2010 18:04:53 +0000 (18:04 +0000)]
* samtools-r818
* for indel calling, do two rounds of probabilistic realignments
Heng Li [Fri, 12 Nov 2010 01:04:07 +0000 (01:04 +0000)]
* samtools-0.1.19-11 (r817)
* only initiate indel calling when 0.2% of reads contain a gap
Heng Li [Thu, 11 Nov 2010 06:22:59 +0000 (06:22 +0000)]
* samtools-0.1.9-10 (r816)
* I know why the forward method fails. it is because of zero base
qualities. when that is fixed, the forward method seems to give
better results than Viterbi, as it should be. I am tired...
Heng Li [Thu, 11 Nov 2010 05:57:15 +0000 (05:57 +0000)]
effectively revert to the viterbi version. The forward realignment gives too many false positives.
Heng Li [Thu, 11 Nov 2010 05:18:02 +0000 (05:18 +0000)]
* samtools-0.1.9-9 (r810)
* use forward, instead of viterbi, for realignment
* realignment is now quality aware
Heng Li [Thu, 11 Nov 2010 03:45:24 +0000 (03:45 +0000)]
* prepare to replace kaln with kprobaln in realignment
Heng Li [Wed, 10 Nov 2010 22:28:50 +0000 (22:28 +0000)]
fixed a typo
Heng Li [Wed, 10 Nov 2010 21:54:46 +0000 (21:54 +0000)]
use zlib for direct reading when BCF_LITE is in use
Heng Li [Wed, 10 Nov 2010 21:32:13 +0000 (21:32 +0000)]
* do not use reads containing too many mismatches for indel calling
* fixed a trivial bug in case of multi-allelic indels
Heng Li [Wed, 10 Nov 2010 18:23:02 +0000 (18:23 +0000)]
* samtools-0.1.9-8 (r809)
* fixed a bug in the indel caller
Heng Li [Wed, 10 Nov 2010 17:24:10 +0000 (17:24 +0000)]
minor change to makefile
Heng Li [Wed, 10 Nov 2010 17:10:21 +0000 (17:10 +0000)]
* samtools-0.1.9-8 (r807)
* collect indel candidates only from specified platforms (@RG-PL)
* merge varFilter and filter4vcf in vcfutils.pl
Heng Li [Wed, 10 Nov 2010 03:05:46 +0000 (03:05 +0000)]
bcftools: compute equal-tail (Bayesian) credible interval
Heng Li [Tue, 9 Nov 2010 21:28:39 +0000 (21:28 +0000)]
added a double-hit filter to avoid overestimated indel likelihood
Heng Li [Tue, 9 Nov 2010 19:12:06 +0000 (19:12 +0000)]
* samtools-0.1.9-7 (r804)
* fixed a bug in the gap caller
Heng Li [Tue, 9 Nov 2010 15:45:33 +0000 (15:45 +0000)]
* samtools-0.1.9-6 (r803)
* mpileup: apply homopolymer correction when calculating GL, instead of before
* bcftools: apply a different prior to indels
Heng Li [Tue, 9 Nov 2010 04:53:15 +0000 (04:53 +0000)]
* samtools-0.1.9-5 (r802)
* relax tandem penalty. this will be made a command-line option in future.
Heng Li [Tue, 9 Nov 2010 04:35:52 +0000 (04:35 +0000)]
* samtools-0.1.9-4 (r801)
* fixed a minor issue in printing indel VCF
Heng Li [Mon, 8 Nov 2010 20:28:14 +0000 (20:28 +0000)]
fixed another silly bug in mpileup's indel caller
Heng Li [Mon, 8 Nov 2010 19:28:27 +0000 (19:28 +0000)]
fixed a silly bug in the indel caller
Heng Li [Mon, 8 Nov 2010 19:07:33 +0000 (19:07 +0000)]
Incorporate patches by Marcel Martin for read counting.
Heng Li [Mon, 8 Nov 2010 18:39:52 +0000 (18:39 +0000)]
* samtools-0.1.9-2 (r797)
* mpileup: indel calling seems to be working
Heng Li [Mon, 8 Nov 2010 15:54:46 +0000 (15:54 +0000)]
indel calling is apparently working, but more information needs to be collected
Heng Li [Mon, 8 Nov 2010 05:39:18 +0000 (05:39 +0000)]
fixed a few bugs in the indel caller. Probably there are more.
Heng Li [Mon, 8 Nov 2010 03:23:16 +0000 (03:23 +0000)]
prepare for the indel caller. It is not ready yet.
Heng Li [Fri, 5 Nov 2010 15:28:23 +0000 (15:28 +0000)]
Revert to r790. The recent changes are not good...
Heng Li [Fri, 5 Nov 2010 04:19:14 +0000 (04:19 +0000)]
* this revision is UNSTABLE
* indel caller seems working, but it is very insensitive and has
several things I do not quite understand.
Heng Li [Fri, 5 Nov 2010 02:58:43 +0000 (02:58 +0000)]
for backup. no effective changes
Heng Li [Wed, 3 Nov 2010 19:51:24 +0000 (19:51 +0000)]
fixed a minor problem in the example coming with kprobaln.c
Heng Li [Tue, 2 Nov 2010 19:41:27 +0000 (19:41 +0000)]
Separate kaln and kprobaln as I am preparing further changes. At
present, the results should be identical to the previous.
Petr Danecek [Tue, 2 Nov 2010 16:19:04 +0000 (16:19 +0000)]
Added -b option: read file names from a file
Heng Li [Sat, 30 Oct 2010 03:17:22 +0000 (03:17 +0000)]
* samtools-0.1.9-2 (r787)
* Allow to set a maximum per-sample depth to reduce memory. However,
BAQ computation is still applied to every read. The speed is not
improved.
Heng Li [Fri, 29 Oct 2010 16:10:40 +0000 (16:10 +0000)]
* samtools-0.1.9-1 (r786)
* samtools: optionally perform exact test for each sample
Heng Li [Fri, 29 Oct 2010 13:42:25 +0000 (13:42 +0000)]
Optionally output "DP", the individual read depth
Heng Li [Thu, 28 Oct 2010 03:10:27 +0000 (03:10 +0000)]
acknowledge Petr and John who have greatly contributed to the project.
Heng Li [Thu, 28 Oct 2010 02:47:47 +0000 (02:47 +0000)]
Release samtools-0.1.9 (r783)
Heng Li [Wed, 27 Oct 2010 23:58:54 +0000 (23:58 +0000)]
fixed a silly bug in pileup
Heng Li [Wed, 27 Oct 2010 18:39:48 +0000 (18:39 +0000)]
* samtools-0.1.8-22 (r781)
* made BAQ the default behavior of mpileup
* updated manual
* in merge, force to exit given inconsistent header when "-R" is not in use.
Heng Li [Wed, 27 Oct 2010 15:01:11 +0000 (15:01 +0000)]
* samtools-0.1.8-21 (r780)
* minor speedup to pileup
Heng Li [Wed, 27 Oct 2010 13:58:56 +0000 (13:58 +0000)]
improve pileup a little bit
Heng Li [Wed, 27 Oct 2010 04:14:43 +0000 (04:14 +0000)]
* samtools-0.1.8-20 (r778)
* speed up pileup, although I do not know how much is the improvement
Heng Li [Tue, 26 Oct 2010 21:26:04 +0000 (21:26 +0000)]
* samtools-0.1.8-19 (r777)
* integrate mpileup features to pileup: min_baseQ, capQ, prob_realn, paired-only and biased prior
Heng Li [Tue, 26 Oct 2010 19:27:46 +0000 (19:27 +0000)]
remove local realignment (probabilistic realignment is still there)
On behalf of John Marshall [Thu, 21 Oct 2010 10:52:38 +0000 (10:52 +0000)]
Add the relevant filename or region to error messages, and cause a failure
exit status where appropriate. Based on a patch provided by Marcel Martin.
Heng Li [Tue, 19 Oct 2010 23:44:31 +0000 (23:44 +0000)]
* Minor code changes. No real effect.
* change quality to 30 in toy.sam
Heng Li [Tue, 19 Oct 2010 03:40:13 +0000 (03:40 +0000)]
added another toy example
Heng Li [Thu, 14 Oct 2010 03:32:12 +0000 (03:32 +0000)]
improve the LD statistics
Heng Li [Wed, 13 Oct 2010 03:49:26 +0000 (03:49 +0000)]
* a minor fix to the -L option
* add ldstats to vcfutils.pl
Heng Li [Tue, 12 Oct 2010 19:51:57 +0000 (19:51 +0000)]
a minor change
Heng Li [Tue, 12 Oct 2010 19:49:06 +0000 (19:49 +0000)]
forget to add the key file
Heng Li [Tue, 12 Oct 2010 19:48:46 +0000 (19:48 +0000)]
* vcfutils.pl: fixed a typo in help message
* added APIs: bcf_append_info() and bcf_cpy()
* calculate adjacent LD
Heng Li [Mon, 11 Oct 2010 15:06:40 +0000 (15:06 +0000)]
added filter for samtools/bcftools genetated VCFs
Heng Li [Tue, 5 Oct 2010 18:05:18 +0000 (18:05 +0000)]
* removed a comment line in kaln.c
* vcfutils.pl fillac works when GT is not the first field
Petr Danecek [Tue, 5 Oct 2010 12:59:36 +0000 (12:59 +0000)]
Convert VCF output of "bcftools view -bgcv" to a valid VCF file
Heng Li [Sun, 3 Oct 2010 02:51:03 +0000 (02:51 +0000)]
* samtools-0.1.8-18 (r763)
* added bcftools manual page
* minor fix to mpileup and view command lines
Heng Li [Sun, 3 Oct 2010 01:46:25 +0000 (01:46 +0000)]
* vcfutils.pl qstats: calculate marginal ts/tv
* allow to call genotypes at variant sites
Heng Li [Fri, 1 Oct 2010 04:29:55 +0000 (04:29 +0000)]
I am changing the gap open probability back to 0.001. It seems that
being conservative here is a good thing...
Heng Li [Fri, 1 Oct 2010 04:11:27 +0000 (04:11 +0000)]
* samtools-0.1.8-17 (r760)
* the default gap open penalty is too small (a typo)
* added comments on hmm_realn
* Java implementation
Heng Li [Thu, 30 Sep 2010 14:12:54 +0000 (14:12 +0000)]
mark samtools-0.1.8-16 (r759)
Heng Li [Thu, 30 Sep 2010 14:12:02 +0000 (14:12 +0000)]
round to the nearest integer
Heng Li [Tue, 28 Sep 2010 21:16:43 +0000 (21:16 +0000)]
I was trying to accelerate ka_prob_glocal() as this will be the
bottleneck. After an hour, the only gain is to change division to
multiplication. OK. I will stop.
Heng Li [Tue, 28 Sep 2010 20:57:49 +0000 (20:57 +0000)]
this is interesting. multiplication is much faster than division, at least on my Mac
Heng Li [Tue, 28 Sep 2010 20:19:13 +0000 (20:19 +0000)]
minor changes
Heng Li [Tue, 28 Sep 2010 19:44:16 +0000 (19:44 +0000)]
prob_realn() seems working!
Heng Li [Tue, 28 Sep 2010 16:48:23 +0000 (16:48 +0000)]
minor
Heng Li [Tue, 28 Sep 2010 16:47:41 +0000 (16:47 +0000)]
Convert phredQ to probabilities
Heng Li [Tue, 28 Sep 2010 16:32:08 +0000 (16:32 +0000)]
Implement the glocal HMM; discard the extention HMM
Heng Li [Tue, 28 Sep 2010 04:06:11 +0000 (04:06 +0000)]
improve numerical stability
Heng Li [Tue, 28 Sep 2010 03:27:54 +0000 (03:27 +0000)]
more comments
Heng Li [Tue, 28 Sep 2010 03:17:16 +0000 (03:17 +0000)]
fixed a bug in banded DP
Heng Li [Tue, 28 Sep 2010 03:05:12 +0000 (03:05 +0000)]
* fixed that weird issue.
* the banded version is NOT working
Heng Li [Tue, 28 Sep 2010 02:57:05 +0000 (02:57 +0000)]
More comments. This version seems working, but something is a little weird...
Heng Li [Mon, 27 Sep 2010 21:21:40 +0000 (21:21 +0000)]
A little code cleanup. Now the forward and backback algorithms give
nearly identical P(x), which means both are close to the correct
forms. However, I have only tested on toy examples. Minor errors in
the implementation may not be obvious.
Heng Li [Mon, 27 Sep 2010 20:55:15 +0000 (20:55 +0000)]
...
On behalf of John Marshall [Mon, 27 Sep 2010 12:19:06 +0000 (12:19 +0000)]
Abort if merge -h's INH.SAM cannot be opened, just as we abort
if any of the IN#.BAM input files cannot be opened.
Also propagate any error indication returned by bam_merge_core()
to samtools merge's exit status.
On behalf of John Marshall [Fri, 24 Sep 2010 15:08:24 +0000 (15:08 +0000)]
Use bam_validate1() to detect garbage records in the event of a corrupt
BAI index file that causes a bam_seek() to an invalid position. At most
one record (namely, the bam_iter_read terminator) is tested per bam_fetch()
call, so the cost is insignificant in the normal case.
On behalf of John Marshall [Fri, 24 Sep 2010 15:00:19 +0000 (15:00 +0000)]
Add bam_validate1().
Heng Li [Wed, 22 Sep 2010 16:07:50 +0000 (16:07 +0000)]
* samtools-0.1.8-15 (r379)
* allow to change capQ parameter in calmd
On behalf of John Marshall [Wed, 22 Sep 2010 15:15:33 +0000 (15:15 +0000)]
When bam_read1() returns an error (return value <= -2), propagate that error
to bam_iter_read()'s own return value. Similarly, also propagate it up to
bam_fetch()'s return value. Previously bam_fetch() always returned 0, and
callers ignored its return value anyway. With this change, 0 continues to
indicate success, while <= -2 (which can be written as < 0, as -1 is never
returned) indicates corrupted input.
bam_iter_read() ought also to propagate errors returned by bam_seek().
main_samview() can now print an error message and fail when bam_fetch()
detects that a .bai index file is corrupted or otherwise does not correspond
to the .bam file it is being used with.
On behalf of John Marshall [Wed, 22 Sep 2010 14:47:42 +0000 (14:47 +0000)]
0 is a successful return value from bam_read1(). (In practice, it never
returns 0 anyway; but all the other callers treat 0 as successful.)
Heng Li [Mon, 20 Sep 2010 21:43:08 +0000 (21:43 +0000)]
* merge files region-by-region. work on small examples but more tests are needed.
Heng Li [Mon, 20 Sep 2010 20:56:24 +0000 (20:56 +0000)]
improve qstats by checking the alleles as well
Heng Li [Fri, 17 Sep 2010 22:12:13 +0000 (22:12 +0000)]
convert UCSC SNP SQL dump to VCF
Heng Li [Fri, 17 Sep 2010 17:02:11 +0000 (17:02 +0000)]
hapmap2vcf convertor
Heng Li [Fri, 17 Sep 2010 14:11:37 +0000 (14:11 +0000)]
* added comments
* VCF->BCF is not possible without knowing the sequence dictionary before hand...
Heng Li [Fri, 17 Sep 2010 13:15:53 +0000 (13:15 +0000)]
* put n_smpl to "bcf1_t" to simplify API a little
Heng Li [Fri, 17 Sep 2010 01:36:01 +0000 (01:36 +0000)]
fixed a bug in indexing
Heng Li [Thu, 16 Sep 2010 20:54:48 +0000 (20:54 +0000)]
* fixed a bug in capQ
* valgrind identifies a use of uninitialised value, but I have not fixed it.
Heng Li [Thu, 16 Sep 2010 19:03:59 +0000 (19:03 +0000)]
* fixed a bug in razip: -c will delete the input file
* copy tabix/bgzip to here
Heng Li [Thu, 16 Sep 2010 17:45:49 +0000 (17:45 +0000)]
* samtools-0.1.8-14 (r727)
* allow to change the capQ parameter at the command line
Heng Li [Thu, 16 Sep 2010 17:38:43 +0000 (17:38 +0000)]
* added varFilter to vcfutils.pl
* reimplement realn(). now it performs a local alignment
* added cap_mapQ() to cap mapping quality when there are many substitutions
Heng Li [Wed, 15 Sep 2010 04:18:31 +0000 (04:18 +0000)]
* convert BCF to QCALL input
Heng Li [Wed, 15 Sep 2010 02:41:50 +0000 (02:41 +0000)]
dynamic band width in realignment
Heng Li [Wed, 15 Sep 2010 02:05:32 +0000 (02:05 +0000)]
fixed a bug in realignment
Heng Li [Wed, 15 Sep 2010 00:54:09 +0000 (00:54 +0000)]
fixed a minor issue
Heng Li [Tue, 14 Sep 2010 23:25:10 +0000 (23:25 +0000)]
fixed a bug in realignment