Hi, Hong!
I just examined the source code to see where
the error happens and what it means.
The customTrack loader is calling wigEncode.
wigEncode checks the coordinates of chromStart and chromEnd
(columns 2 and 3 in many input formats)
by calling sqlLongLong (or sometimes sqlSigned) in lib/sqlNum.c.
Perhaps the error message would be more specific if it said
"invalid signed integer" instead of "invalid signed number".
The error message prints out the input exactly as it is,
i.e. the text itself and not a number or numeric interpretation,
so it seems very likely that "1.13e+08" does appear in the input.
This sqlLongLong routine reads a piece of ascii text
and translates it to a long long, or it gives an error.
It only tolerates an optional leading minus sign [-] and then
the digits [0-9]. Periods, pluses and the letter e are not allowed.
Can you please check your input wigAscii text again?
Can you send a few lines from the top of your customTrack
file so I can see exactly what columns of data you have?
Thanks!
FYI from file lib/sqlNum.c:
long long sqlLongLong ( char * s )
{
long long res = 0;
char *p, *p0 = s;
if (*p0 == '-')
p0++;
p = p0;
while ((*p >= '0') && (*p <= '9'))
{
res *= 10;
res += *p - '0';
p++;
}
/* test for invalid character, empty, or just a minus */
if ((*p != '\0') || (p == p0))
errAbort("invalid signed number: \"%s\"", s);
if (*s == '-')
return -res;
else
return res;
}
-Galt
Ar 5/18/2010 12:16 PM, scríobh Jennifer Jackson:
> Hi Hong,
>
> Thanks for the feedback. Good to know the scientific notation is not in
> the data.
>
> There are a few things to check:
>
> 1) Are you loading data through the custom track submission form? The
> error you sent before looked like a regular "load track" error for users
> with a mirror. If you want to load as a custom track, with the custom
> track headers, you must use the custom track submission form.
>
> 2) There are a few problems with the track line. Clearing these up will
> likely help.
>
> autoScale="off" (remove the "")
> graphType="bar" (remove the "")
>
> http://genome.ucsc.edu/goldenPath/help/hgWiggleTrackHelp.html
> http://genome.ucsc.edu/goldenPath/help/wiggle.html (scroll to "Data Values"
> http://genome.ucsc.edu/goldenPath/help/hgTracksHelp.html#TRACK
> http://hgwdev.cse.ucsc.edu/~kent/src/unzipped/hg/makeDb/trackDb/ (see
> sections THE TYPE FIELD, #14 and OTHER FIELDS).
>
> 3) Using an alternate data format would likely be much better given the
> data size. Track line attributes for .wig also work for .bigWig. The
> ".wig" data type is being retired anyway - the recommended replacement
> is bigWig. See this document for the graphing data types including many
> links to associated help docs:
> http://genomewiki.ucsc.edu/index.php/Selecting_a_graphing_track_data_format
> http://genome.ucsc.edu/goldenPath/help/bigWig.html
>
> Conversion program wigToBigWig in the kent source tree or available via
> ftp pre-compiled here (for some systems):
> http://hgdownload.cse.ucsc.edu/admin/exe/
>
> Hopefully this is helpful. Please let us know if you continue to have
> problems,
> Jennifer
>
> ---------------------------------
> Jennifer Jackson
> UCSC Genome Informatics Group
> http://genome.ucsc.edu/
>
> On 5/18/10 11:44 AM, Hong Ji wrote:
>> Hi there,
>> I loaded my track as a custom track and I added the following header:
>> track type=wiggle_0 name="XXX" description="YYY" maxHeightPixels=200:100:40
>> viewLimits=-5.0:5.0 autoScale="off" graphType="bar" visibility=full
>> color=250,0,0 altColor=0,100,200 priority=19
>>
>> I have check the list and didn't find any scientific notation. I do have
>> more than 200,000 rows in my txt file, will that cause any problem?
>>
>> Hong
>>
>> Hong Ji, Ph. D
>> Postdoc Fellow
>> Center for Epigenetics
>> Johns Hopkins School of Medicine
>> 580 Rangos Building,
>> 855 North Wolfe Street
>> Baltimore, MD 21205
>> Email:[email protected]
>> Lab phone: 410-614-3478
>>
>>
>>
>> ----- Original Message -----
>> From: Jennifer Jackson<[email protected]>
>> Date: Monday, May 17, 2010 6:34 pm
>> Subject: Re: [Genome] error
>> To: Hong Ji<[email protected]>
>> Cc: [email protected]
>>
>>> Update:
>>>
>>> If by chance you are using scientific notation in the input file, that
>>> may be the root cause of your problems. Try changing those values to
>>> be fully expanded integers then reload.
>>>
>>> As a side note, we will be making the exit more informative when a
>>> non-integer is expected but not in the input. This will be made
>>> available in the next code base release at the latest. This change
>>> will have no impact for actual usage (expanded integers still will be
>>> required), but it should help make the exit less cryptic.
>>>
>>> Thank you for letting us know about this and we hope that one of these
>>> options resolves your loading issues,
>>>
>>> Thanks,
>>> Jennifer
>>>
>>> On 5/17/10 3:15 PM, Jennifer Jackson wrote:
>>>> Hello,
>>>>
>>>> The name of your file looks like a custom track, but you are loading
>>> as
>>>> a native track into your own mirror? Did you remove all of the
>>>> browser/track lines from the file before loading?
>>>>
>>>> If that is not the problem, please send more details about your exact
>>>> procedure (assembly, your mirror or genome.ucsc?, load as native track
>>>> or custom track, small data sample, exact command line arguments - if
>>>> any). We will try to diagnose the problem from there.
>>>>
>>>> Thanks,
>>>> Jennifer
>>>>
>>>> ---------------------------------
>>>> Jennifer Jackson
>>>> UCSC Genome Informatics Group
>>>>
>>>>
>>>> On 5/15/10 9:00 AM, Hong Ji wrote:
>>>>> I am trying to upload a genome track and keep getting the following
>>> error:
>>>>>
>>>>> track load error (track name='ct_xxx_6191'):
>>>>> invalid signed number: "1.13e+08"
>>>>>
>>>>> I tried to upload the top few lines and it worked. Anyone know
>>> what’s wrong? Thanks!
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> Genome maillist - [email protected]
>>>>>
>>>> _______________________________________________
>>>> Genome maillist - [email protected]
>>>>
> _______________________________________________
> Genome maillist - [email protected]
> https://lists.soe.ucsc.edu/mailman/listinfo/genome
_______________________________________________
Genome maillist - [email protected]
https://lists.soe.ucsc.edu/mailman/listinfo/genome
