Add new annot-tsv program #1619

pd3 · 2023-05-18T14:43:02Z

See also
https://raw.githubusercontent.com/pd3/utils/master/annot-regs/doc.annot-regs.pdf
https://github.com/pd3/utils/tree/master/annot-regs

pd3 · 2023-05-30T08:40:49Z

Mmm, I am confused why the tests complain about strdup being implicitly declared, even after #include <string.h> has been added, while tabix code gets away with it...?

jkbonfield · 2023-05-30T15:41:48Z

In subsequent emails I think I decided on better terminology of "input" and "annotation". "Source" still feels nebulous and not very descriptive of the function.

The usage has:

   -s, --source-file FILE          Source file to take annotations from
   -t, --target-file FILE          Target file to be extend with annotations from -s

From this, it sounds like "target" is our file we are extending, with annotations from "source".

The man page says:

       -f, --transfer SRC:TGT
           Comma-separated list of columns to transfer. If the SRC column does not exist,
           interpret it as the default value to fill in when a match is found or a dot (".") when
           a match is not found. If the TGT column does not exist, a new column is created. If
           the TGT column already exists, its values will be overwritten when overlap is found
           and left as is otherwise.

This is a bit ambiguous as it sort of implies it can work in either direction.

I'm still struggling to work out if it's completely symmetric and we basically have file1 file2 that are merged (akin to unix join), or whether we have a primary input file and an annotation file that is used to modify the input. If so, which is actually the primary input? I think this all hinges on the "If the SRC column does not exist" vs "If the TGT column does not exist" interpretation. They're not equal, so clearly one is more important than the other and one of the two files is the primary file and the other is the secondary file, but I find the flow of data hard to grok.

Think of it in this way. Let's say we have an awk script in a very strict layout:

$1 ~ /foo/ {$2="abc"}
$1 ~ /bar/ {$2="pqr"}
$1 ~ /ram/ {$2="xyz"}
{print}

We can feed any data to it and the script will amend specific matching columns based on the contents of other columns. It's clear here that a null script degenerates to "cat". It's clear the input is the input, and the script is the script. We never really do things like pipe a script into awk and specify a filename as input on the command line.

Here it's a little different as we have to have at least one matching column between two files (hence why I say it's like a more powerful "join"). That does change things. I don't understand the ordering though. What's the use case for it working on unordered data? What's the benefit here? It just feels like it'd make it far slower than necessary. Or the half way house: is there a benefit to one order (primary) and one unordered (annotation) file?

I'm pleased however to see the addition of "-o" for output since I last looked at this, and renaming the old overlap option.

pd3 · 2023-05-31T06:52:36Z

In subsequent emails I think I decided on better terminology of "input" and "annotation". "Source" still feels nebulous and not very descriptive of the function.

Unless there is a danger of confusion, let's preserve the existing naming please. It is to avoid conflicting short names, I already had to make sacrifices by abandoning the term "destination" which you disliked.

The naming was intended to imply the information flow and for that reason I liked best the original "source" and "destination". Using that naming would also prevent the next question:

This is a bit ambiguous as it sort of implies it can work in either direction.

I'm still struggling to work out if it's completely symmetric and we basically have file1 file2 that are merged (akin to unix join), or whether we have a primary input file and an annotation file that is used to modify the input.

It is not symmetrical and is not intended to imply a change in direction. It aims to explain that the argument to --transfer are column names, in the source and the target file, and what happens if such column does or does not exist in the source and the target file: if the column exists in both, values in the target file are overwritten by values from the source file on match or left as is when there is no match; if such column does not exist in the target file, a new column with that name will be created in the target file; if such column does not exist in the source file, the name is interpreted as a string to be filled in the target file. For example, if the column MATCH does not exist in either file, --transfer MATCH:MATCH (which can be abbreviated to --transfer MATCH) will create a new column MATCH in the target file and put the value MATCH whenever there is a match or a dot when there isn't.

@daviesrob, @whitwham could you please help to improve the wording of the man page and the usage text to make it clearer? Despite my best effort I am not succeeding in making this understandable.

I don't understand the ordering though. What's the use case for it working on unordered data? What's the benefit here? It just feels like it'd make it far slower than necessary. Or the half way house: is there a benefit to one order (primary) and one unordered (annotation) file?

I am confused by the question. There are two types of ordering you might be asking, by column and by row. You seem to be asking about column ordering, which is the only type of ordering mentioned throughout the man page and the code.

To state the obvious, the benefit of supporting tab-delimited files with columns in arbitrary order is that one can use the program on files with columns in arbitrary order. For example, on files that were exported from spreadsheets, people tend to use them a lot. Regarding ordering by row, even though that is not mentioned anywhere, the program can work with data unordered by row as well. That's thanks to the regidx library.

As a side note, comparing the program to awk, join etc is misleading. It is also a bit like grep and bedtools, but mostly provides a unique functionality that is not found elsewhere.

jkbonfield · 2023-05-31T08:27:05Z

I meant row ordering, so classic sorted vs unsorted (or non-position sorted) genomic data.

We have lots of algorithms that are efficient because they require sorted data. Eg samtools merge. Here you seem to be stating that both files can be unsorted. This obviously means it is using a less efficient strategy, which is fine if that is the user requirement. It also depends on the size of files you're designing it to operate on. Is this for small data of a few thousand rows, or are you planning for it to lift over annotations from files with 10s of millions of records?

I'm just asking where that requirement for operating on unsorted data comes from. What inputs are we dealing with that aren't sorted? (Even unix "join" requires sorted inputs)

jkbonfield · 2023-05-31T08:30:32Z

In subsequent emails I think I decided on better terminology of "input" and "annotation". "Source" still feels nebulous and not very descriptive of the function.

Unless there is a danger of confusion, let's preserve the existing naming please. It is to avoid conflicting short names, I already had to make sacrifices by abandoning the term "destination" which you disliked.

Maybe so, but I asked others in the NPG group where they thought the input came from and where the output went, and universally they said source and destination! Let's just agree to disagree on this one. I'm looking for better terms, not to rehash old arguments.

I still think input and output are by far the most universally understood terms. As you've agreed it's not symmetric and one file is the primary input, then why not call that file input, and the other something more appropriate to its function?

jkbonfield · 2023-05-31T10:18:17Z

With a trivial src and dst (transfer) file:

$ cat /tmp/src.1.txt
#chr	beg	end	tag3
1	14	15	a
1	35	35	b

$ cat /tmp/dst.1.txt
#chr	beg	end	tag1	tag2
1	14	15	x1	x2
1	35	35	y1	y2

I can do a nop function of specifying nothing to transfer over:

$ ./annot-tsv -s /tmp/src.1.txt -t /tmp/dst.1.txt  -c chr,beg,end
#[1]chr	[2]beg	[3]end	[4]tag1	[5]tag2
1	14	15	x1	x2
1	35	35	y1	y2

I notice a few things. Firstly, the header line has changed format. Is that deliberate or just debugging left over? It's undocumented.

Also what you label as the dst file in your tests (old destination) is the default output, and not the source as I had expected! If we use -f to transfer tags, then these come from source and not the target file. This is still confusingly named therefore. PLEASE consider just renaming the input to the "input" as it's then crystal clear what it is.

I also tried stress testing. Removing one of the columns from one src row (I deleted the "b" at the end of the 2nd line) to get illegal data caused a core dump:

./annot-tsv -s /tmp/src.1.txt -t /tmp/dst.1.txt -f tag3
AddressSanitizer:DEADLYSIGNAL
=================================================================
==8978==ERROR: AddressSanitizer: SEGV on unknown address (pc 0x0000004d3015 bp 0x7fff4ad0c9d0 sp 0x7fff4ad0c760 T0)
==8978==The signal is caused by a READ memory access.
==8978==Hint: this fault was caused by a dereference of a high value address (see register values below).  Disassemble the provided pc to learn which register was used.
    #0 0x4d3015 in __ac_X31_hash_string /nfs/users/nfs_j/jkb/work/samtools_master/htslib/./htslib/khash.h:401
    #1 0x4d3015 in kh_get_str2int /nfs/users/nfs_j/jkb/work/samtools_master/htslib/./htslib/khash_str2int.h:30
    #2 0x4d3015 in khash_str2int_has_key /nfs/users/nfs_j/jkb/work/samtools_master/htslib/./htslib/khash_str2int.h:69
    #3 0x4d3015 in process_line /nfs/users/nfs_j/jkb/work/samtools_master/htslib/annot-tsv.c:685
    #4 0x4d596a in main /nfs/users/nfs_j/jkb/work/samtools_master/htslib/annot-tsv.c:875
    #5 0x7f8fdbee3c86 in __libc_start_main /build/glibc-CVJwZb/glibc-2.27/csu/../csu/libc-start.c:310
    #6 0x41d939 in _start ??:?

AddressSanitizer can not provide additional info.
SUMMARY: AddressSanitizer: SEGV /nfs/users/nfs_j/jkb/work/samtools_master/htslib/./htslib/khash.h:401 in __ac_X31_hash_string
==8978==ABORTING
Aborted

We need to stress test more corner cases so it's robust.

jkbonfield · 2023-05-31T10:29:49Z

What specifies the ordering?

My source file has columns "tag3" and "m" in that order, but I always get them in the opposite order regardless of what I specify in the -f option.

@ seq4d[samtools.../htslib]148; cat /tmp/src.1.txt
#chr	beg	end	tag3	m
1	14	15	a	1
1	35	35	b	01
@ seq4d[samtools.../htslib]; ./annot-tsv -s /tmp/src.1.txt -t /tmp/dst.1.txt -f m,tag3
#[1]chr	[2]beg	[3]end	[4]tag1	[5]tag2	[6]m	[7]tag3
1	14	15	x1	x2	1	a
1	35	35	y1	y2	01	b
@ seq4d[samtools.../htslib]; ./annot-tsv -s /tmp/src.1.txt -t /tmp/dst.1.txt -f tag3,m
#[1]chr	[2]beg	[3]end	[4]tag1	[5]tag2	[6]m	[7]tag3
1	14	15	x1	x2	1	a
1	35	35	y1	y2	01	b

Also, please consider making the usage appear to stdout unless an error has occurred. This is already a cause of a lot of "known failures" in the Samtools test harness, and we shouldn't be adding more here. Specifically an error is an error (eg anno-tsv --unknown-opt), but a request for usage (anno-tsv -h) is not. Same with exit codes. -h should have exit code 255. (Nor should errors really - you probably wanted exit(1) instead of exit(-1) there)

pd3 · 2023-05-31T11:32:11Z

I can do a nop function of specifying nothing to transfer over:

In this mode it functions as grep would.

pd3 · 2023-05-31T11:34:14Z

I notice a few things. Firstly, the header line has changed format. Is that deliberate or just debugging left over? It's undocumented.

That's a deliberate feature. If it is bothersome to anyone, we can add an option to not modify the header. I would leave this for future improvements though.

pd3 · 2023-05-31T11:40:13Z

Also what you label as the dst file in your tests (old destination) is the default output, and not the source as I had expected! If we use -f to transfer tags, then these come from source and not the target file. This is still confusingly named therefore. PLEASE consider just renaming the input to the "input" as it's then crystal clear what it is.

I really don't understand the problem. The documentation is clear on that, no? The program transfers columns from the file with source annotations (-s) to the destination file, newly "target" file (-t).

  -s, --source-file FILE          Source file to take annotations from
  -t, --target-file FILE          Target file to be extend with annotations from -s

To turn this around: both files are technically inputs and typically both have annotations. I can be equally confused here.

Fixes samtools#1619 (comment)

pd3 · 2023-05-31T11:49:07Z

./annot-tsv -s /tmp/src.1.txt -t /tmp/dst.1.txt -f tag3
AddressSanitizer:DEADLYSIGNAL
=================================================================
==8978==ERROR: AddressSanitizer: SEGV on unknown address (pc 0x0000004d3015 bp 0x7fff4ad0c9d0 sp 0x7fff4ad0c760 T0)
==8978==The signal is caused by a READ memory access.
==8978==Hint: this fault was caused by a dereference of a high value address (see register values below).  Disassemble the provided pc to learn which register was used.
    #0 0x4d3015 in __ac_X31_hash_string /nfs/users/nfs_j/jkb/work/samtools_master/htslib/./htslib/khash.h:401
    #1 0x4d3015 in kh_get_str2int /nfs/users/nfs_j/jkb/work/samtools_master/htslib/./htslib/khash_str2int.h:30
    #2 0x4d3015 in khash_str2int_has_key /nfs/users/nfs_j/jkb/work/samtools_master/htslib/./htslib/khash_str2int.h:69
    #3 0x4d3015 in process_line /nfs/users/nfs_j/jkb/work/samtools_master/htslib/annot-tsv.c:685
    #4 0x4d596a in main /nfs/users/nfs_j/jkb/work/samtools_master/htslib/annot-tsv.c:875
    #5 0x7f8fdbee3c86 in __libc_start_main /build/glibc-CVJwZb/glibc-2.27/csu/../csu/libc-start.c:310
    #6 0x41d939 in _start ??:?

AddressSanitizer can not provide additional info.
SUMMARY: AddressSanitizer: SEGV /nfs/users/nfs_j/jkb/work/samtools_master/htslib/./htslib/khash.h:401 in __ac_X31_hash_string
==8978==ABORTING
Aborted

Thank you, added a commit to address issues like this ee299e3. More breaking test cases will be much appreciated

pd3 · 2023-05-31T11:56:19Z

What specifies the ordering?

My source file has columns "tag3" and "m" in that order, but I always get them in the opposite order regardless of what I specify in the -f option.

New columns are only appended, never added among existing ones. I suspect the dst.1.txt file in this example contains the column m already? Otherwise in my tests the ordering of new columns can be influenced by giving them in different order in the -f option.

Reordering of existing columns in the target/destination file is not considered in this version, perhaps it could be addressed in a future enhancement request.

pd3 · 2023-05-31T12:33:26Z

Also, please consider making the usage appear to stdout unless an error has occurred. This is already a cause of a lot of "known failures" in the Samtools test harness, and we shouldn't be adding more here. Specifically an error is an error (eg anno-tsv --unknown-opt), but a request for usage (anno-tsv -h) is not. Same with exit codes. -h should have exit code 255. (Nor should errors really - you probably wanted exit(1) instead of exit(-1) there)

The exit code for -h was now changed to EXIT_SUCCESS and EXIT_FAILURE otherwise, 1424ca6

daviesrob · 2023-06-27T11:08:39Z

annot-tsv.c

+#define NBP_IS_END(x)  (((x)&1)==1)
+typedef struct
+{
+    int n,m;            // n is a multiple of two: breakpoints are stored in regs, not regions


These should be size_t.

int or uint should be sufficient for any normal use. But ok, no harm in using size_t.

daviesrob · 2023-06-27T11:09:08Z

annot-tsv.c

+typedef struct
+{
+    int n,m;            // n is a multiple of two: breakpoints are stored in regs, not regions
+    hts_pos_t *regs;     // change to uint64_t for very large genomes


hts_pos_t is uint64_t.

The original annot-regs used int, the comment was not updated when transitioning to annot-tsv..

daviesrob · 2023-06-27T11:13:48Z

annot-tsv.c

+}
+static inline void nbp_add(nbp_t *nbp, hts_pos_t beg, hts_pos_t end)
+{
+    if ( end >= REGIDX_MAX>>1 ) error("Error: the coordinate is too big (%u). Possible todo: switch to uint64_t\n",end);


REGIDX_MAX is (1ULL << 35) so already only fits in 64 bits.

annot-tsv.c

daviesrob · 2023-06-27T11:21:24Z

annot-tsv.c

+
+typedef struct
+{
+    uint32_t n,m;


These should really be size_t too...

This would mean a file with more than 4,294,967,295 columns

daviesrob · 2023-06-27T11:30:02Z

annot-tsv.c

+static int read_next_line(dat_t *dat)
+{
+    if ( dat->line.l ) return dat->line.l;
+    if ( hts_getline(dat->fp, KS_SEP_LINE, &dat->line) > 0 ) return dat->line.l;


You need to check for errors reported by hts_getline() (i.e. return value < -1) to tell them apart from EOF.

daviesrob · 2023-06-27T11:42:08Z

annot-tsv.c

+    char *temp_dir, *out_fname;
+    BGZF *out_fp;
+    int allow_dups, reciprocal, ignore_headers, max_annots, mode;
+    float overlap;


Why use float here instead of double?

The precision of double is overkill here, but happy to change...

daviesrob · 2023-06-27T12:38:38Z

annot-tsv.c

+    if (a > b) return 1;
+    return 0;
+}
+static int nbp_length(nbp_t *nbp)


Should return hts_pos_t

daviesrob · 2023-06-27T12:51:08Z

annot-tsv.txt

I'd prefer to only have one copy of the manual page checked in - and preferably the nroff one so HTSlib doesn't gain a build dependency on asciidoc.

Are you suggesting that the man page is formatted in nroff directly? @jkbonfield originally suggested text markdown is fine. Maybe we should discuss offline.

I agree with the reserveration to asciidoc though, asciidoctor is what we've been using in bcftools

I think I said it's fine in principle, but if we use it it should be universal and not just one thing. Adding dependencies without consideration for other devs and users isn't ideal so we have to be considerate and not just drift into policy decisions like this.

Bcftools aleady uses this, so it's fine there. Htslib right now is nroff -man so we should stick to that unless we decide to move enmasse to a new regime.

daviesrob · 2023-06-27T12:53:31Z

annot-tsv.1

+The program was written by Petr Danecek and was originally published on github as annot\-regs
+.SH "COPYING"
+.sp
+The MIT/Expat License or GPL License, see the LICENSE document for details\&. Copyright (c) Genome Research Ltd\&.


MIT/Expat only. GPL is not currently mentioned in the LICENSE document, and I don't see why anyone would want to use it if you can go for MIT/Expat instead.

daviesrob · 2023-08-18T11:32:59Z

test/test.pl

-test_plugin_loading($opts);
-test_realn($opts);
-test_bcf_set_variant_type($opts);
+run_test(\&test_bgzip,$opts, 0);


This change to enable the --function option doesn't quite work, because some of the tests currently rely on artefacts made by earlier tests. Rather than try to fix that as part of this PR, it would be easier to just have the annot-tsv tests run the old way. Adding the --function option can then be done as separate work in a new PR.

I know it is not bullet proof for all tests, but it works if you want to rerun and debug the ones you're currently developing. Saves lot of time and does not have to be used, I'd advocate for keeping it

daviesrob · 2023-08-18T11:41:25Z

test/test.pl

+run_test(\&test_realn,$opts);
+run_test(\&test_bcf_set_variant_type,$opts);
+
+run_test(\&test_annot_tsv,$opts,src=>'src.1.txt',dst=>'dst.1.txt',out=>'out.1.1.txt',args=>'-f smpl:overlap --allow-dups');


To keep in line with the rest of this top-level code, it should just call test_annot_tsv($opts); and then have that run the individual tests.

daviesrob · 2023-08-18T13:09:23Z

test/test.pl

+    my $pid = fork();
+    if ( !$pid )
+    {
+        exec('bash', '-o','pipefail','-c', "($cmd) 2>$tmp.e >$tmp.o");
+    }
+    waitpid($pid,0);


This could just be replaced by

system('bash', '-o','pipefail','-c', "($cmd) 2>$tmp.e >$tmp.o");

Compared with _cmd(), the $ENV{TEST_PRECMD} functionality has been lost (although we don't use it that much these days, anyway).

daviesrob · 2023-08-18T13:13:06Z

test/test.pl

+    my (@out,@err);
+    if ( open(my $fh,'<',"$tmp.o") )
+    {
+        @out = <$fh>;


I'm not sure why we use an array for this (it's in other code too). It would be more efficient to just read everything into a string:

local $/; # Read whole file $out = <$fh>;

This is just out of habit, usually I'd do something with the output per-line. I don't think this is a performance issue in these tests, but okay.

daviesrob · 2023-08-18T13:30:23Z

test/test.pl

@@ -216,7 +304,16 @@ sub test_cmd
        }
        else
        {
-            failed($opts,$test,"The outputs differ:\n\t\t$$opts{path}/$args{out}\n\t\t$$opts{path}/$args{out}.new");
+            if ( exists($args{exp}) && !-e "$$opts{path}/$args{out}" )


The second part of this will always be false due to the condition for the surrounding if statement.

daviesrob · 2023-08-18T13:41:15Z

test/test.pl

-    my ($ret,$out) = _cmd("$args{cmd}");
-    if ( $ret ) { failed($opts,$test); return; }
+    my ($ret,$out,$err) = _cmd3("$args{cmd}");
+    if ( length($err) ) { $err =~ s/\n/\n\t\t/gs; $err = "\n\n\t\t$err\n"; }


Can be done more easily with $err =~ s/^/\t/mg;

daviesrob · 2023-08-18T13:44:35Z

test/test.pl

+                print $fh $exp;
+                close($fh);
+            }
+            my @diff = `diff $$opts{path}/$args{out} $$opts{path}/$args{out}.new`;


Easier to capture into a string, then use s/^/\t/mg.

daviesrob · 2023-08-18T13:49:33Z

test/test.pl

+    waitpid($pid,0);
+
+    my $status  = $? >> 8;
+    my $signal  = $? & 127;


This value is never used.

Fixes samtools#1619 (comment)

See also https://raw.githubusercontent.com/pd3/utils/master/annot-regs/doc.annot-regs.pdf https://github.com/pd3/utils/tree/master/annot-regs

Fixes samtools#1619 (comment)

- checks for memory allocation failures - sanitize numeric types and clean up old comments

Instead of trying to get the function name from a code ref, pass it in directly as a string. It can be looked up in the symbol table to get the corresponding code ref.

Some tests rely on others having been run first. Add a "needed_by" arg to run_test() (essentially an alias for "cmd", but it better documents the relationship) so these dependencies can be resolved. Add "needed_by" args so all tests work when run on their own.

Use local $/; to read entire files into a string. Simplify substitutions to put tabs at the start of lines. Remove unused chunk of code from test_cmd(), and capture diff output in a string instead of an array.

Add HTS_FORMAT, HTS_NORETURN annotations to error() function Make error() flush stdout, stderr so errors are always written after any output. Fix reported error() format string mismatches Fix checks on number of columns

And select the nroff version as the single source of truth.

daviesrob · 2023-10-12T15:51:26Z

Rebased, with some squashing of a couple of trivial bug-fix commits, and some adjustments from me...

Fixes #1619 (comment)

pd3 requested review from jkbonfield and daviesrob May 18, 2023 14:43

jkbonfield removed their request for review May 31, 2023 10:49

pd3 added a commit to pd3/htslib that referenced this pull request May 31, 2023

Provide a nill (dot) value when the field is empty

ee299e3

Fixes samtools#1619 (comment)

daviesrob reviewed Jun 27, 2023

View reviewed changes

samtools deleted a comment from jkbonfield Jun 28, 2023

daviesrob reviewed Aug 18, 2023

View reviewed changes

daviesrob pushed a commit to daviesrob/htslib that referenced this pull request Aug 29, 2023

Provide a nill (dot) value when the field is empty

a43e806

Fixes samtools#1619 (comment)

daviesrob pushed a commit to daviesrob/htslib that referenced this pull request Oct 12, 2023

Provide a nill (dot) value when the field is empty

9563039

Fixes samtools#1619 (comment)

pd3 added 8 commits October 12, 2023 16:32

Add new annot-tsv program

4c8515d

See also https://raw.githubusercontent.com/pd3/utils/master/annot-regs/doc.annot-regs.pdf https://github.com/pd3/utils/tree/master/annot-regs

Output full diff on failing tests

016692f

Output full diff on failing tests

48e7b7c

Make qsort in regidx order-reproducible across platforms

0d36b85

Provide a nill (dot) value when the field is empty

fb15ada

Fixes samtools#1619 (comment)

Use EXIT_SUCCESS with -h and EXIT_FAILURE on errors

e99eca2

Clarify usage, include output example

f1d381c

Address various comments

9ab9e75

- checks for memory allocation failures - sanitize numeric types and clean up old comments

daviesrob added 6 commits October 12, 2023 16:32

Simplify run_test function

434a94a

Instead of trying to get the function name from a code ref, pass it in directly as a string. It can be looked up in the symbol table to get the corresponding code ref.

Simplify test/test.pl

2c3d7b8

Use local $/; to read entire files into a string. Simplify substitutions to put tabs at the start of lines. Remove unused chunk of code from test_cmd(), and capture diff output in a string instead of an array.

Move all annot_tsv tests into their own function

4942a5f

Fix up some annot-tsv error checks / reports

4ac265c

Add HTS_FORMAT, HTS_NORETURN annotations to error() function Make error() flush stdout, stderr so errors are always written after any output. Fix reported error() format string mismatches Fix checks on number of columns

Reformulate man page for house style

c3aeef2

And select the nroff version as the single source of truth.

daviesrob force-pushed the annot-tsv branch from 2fd54e3 to c3aeef2 Compare October 12, 2023 15:49

whitwham self-assigned this Oct 13, 2023

whitwham merged commit 99415e2 into samtools:develop Oct 13, 2023
9 checks passed

whitwham pushed a commit that referenced this pull request Oct 13, 2023

Provide a nill (dot) value when the field is empty

808a380

Fixes #1619 (comment)

pd3 deleted the annot-tsv branch November 30, 2023 10:35

Add new annot-tsv program #1619

Add new annot-tsv program #1619

Conversation

pd3 commented May 18, 2023

pd3 commented May 30, 2023 • edited Loading

jkbonfield commented May 30, 2023

pd3 commented May 31, 2023

jkbonfield commented May 31, 2023 • edited Loading

jkbonfield commented May 31, 2023

jkbonfield commented May 31, 2023

jkbonfield commented May 31, 2023

pd3 commented May 31, 2023

pd3 commented May 31, 2023

pd3 commented May 31, 2023

pd3 commented May 31, 2023

pd3 commented May 31, 2023 • edited Loading

pd3 commented May 31, 2023

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jkbonfield Jun 28, 2023 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

daviesrob commented Oct 12, 2023

pd3 commented May 30, 2023 •

edited

Loading

jkbonfield commented May 31, 2023 •

edited

Loading

pd3 commented May 31, 2023 •

edited

Loading

jkbonfield Jun 28, 2023 •

edited

Loading