[Gmod-gbrowse] Help with gbrowse

Discussion:

Scott Cain

2017-01-23 14:58:34 UTC

Hi Shabari,

I'm cc'ing the GBrowse mailing list so the discussion can be archived.

About the initial landmark: can you provide a copy of the config file that
isn't working? I think I described in an earlier email that the initial
landmark only comes into effect when someone is visiting the site who has
never been there (since after the first visit, a cookie remembers the last
cooridinates visited). If there is some other problem with the initial
landmark please describe the problem more thoroughly.

About the links out: please provide the track configuration that isn't
working and a sample of the GFF that should be affected by the
configuration. It's pretty hard to troubleshoot from your description :-)

Thanks,
Scott

On Mon, Jan 23, 2017 at 1:09 AM, Shabari Subramanian <

Hi Scott,
I recently joined Phoenix Bioinformatics. I have been working on fixing
some bugs on the gbrowse viewer and ran into a wall. Tanya Berardini
suggested i reach out to you.
1. I set the inital landmark on the species conf file in /etc/gbrowse to
point to specific chr location. However the landmark field still comes up
blank on the gbrowse page for the species.
2. I am trying to set a link from gbrowse tracks of ortholgous genes to
the source phytozome pages spcific to the gene on the track. An example of
such a link is https://phytozome.jgi.doe.gov/
pz/portal.html#!gene?organism=Athaliana&searchText=locusName:AT1G01040. I
set this link within the conf file for the species. However, when i mouse
over the track, the link points back to localhost. If i change the link to
http://www.phytozome.net, the link works fine. http links on other tracks
work fine. None of the other links are https links. Any idea why gbrowse is
behaving such.
Thanks for your help.
Best
-shabari

--
------------------------------------------------------------------------
Scott Cain, Ph. D. scott at scottcain dot
net
GMOD Coordinator (http://gmod.org/) 216-392-3087
Ontario Institute for Cancer Research

Scott Cain

2017-01-23 20:22:26 UTC

Permalink

Hi Shabari,

I was a little confused by your original description of the link problem,
which was compounded by the fact that the URL you have in the config looks
like it only works for A. thaliana features, but the GFF features don't
seem to have any (assuming that's what the Note attribute means). When I
tried to manually build a function URL with the callback you provided, I
just got a not found message from Phytozome.

But, all of this is to say that I went back and read your original
description of the problem and it doesn't feel like a problem with the
GBrowse. Is it possible that you have something either in your /etc/hosts
file or at your institution that is rerouting these requests? The
substituting in localhost for the JGI domain is a little weird. Do you
have a publicly available instance of this I can look at?

Thanks,
Scott

On Mon, Jan 23, 2017 at 2:07 PM, Shabari Subramanian <

Hi Scott,
Apologies for not being clearer.
Thank you for clarifying how the initial landmark works. I am clear on the
behavior now.
Please find attached two files, the config file with the broken link to
phytozome and a second file with the gff formatted annotations file for the
gbrowse track in question.
Here is the part of the config file that i added to try to link to
phytozome.
link = sub {
my $feature = shift;
my $Name=$feature->name();
return "https://phytozome.jgi.doe.
gov/pz/portal.html#!gene?organism=Athaliana&searchText=locusName:".$Name;
}
Please let me know if you need more information.
Thanks for your help.
-shabari

Post by Scott Cain
Hi Shabari,
I'm cc'ing the GBrowse mailing list so the discussion can be archived.
About the initial landmark: can you provide a copy of the config file
that isn't working? I think I described in an earlier email that the
initial landmark only comes into effect when someone is visiting the site
who has never been there (since after the first visit, a cookie remembers
the last cooridinates visited). If there is some other problem with the
initial landmark please describe the problem more thoroughly.
About the links out: please provide the track configuration that isn't
working and a sample of the GFF that should be affected by the
configuration. It's pretty hard to troubleshoot from your description :-)
Thanks,
Scott
On Mon, Jan 23, 2017 at 1:09 AM, Shabari Subramanian <

Hi Scott,
I recently joined Phoenix Bioinformatics. I have been working on fixing
some bugs on the gbrowse viewer and ran into a wall. Tanya Berardini
suggested i reach out to you.
1. I set the inital landmark on the species conf file in /etc/gbrowse to
point to specific chr location. However the landmark field still comes up
blank on the gbrowse page for the species.
2. I am trying to set a link from gbrowse tracks of ortholgous genes to
the source phytozome pages spcific to the gene on the track. An example of
such a link is https://phytozome.jgi.doe.gov/
pz/portal.html#!gene?organism=Athaliana&searchText=locusName:AT1G01040.
I set this link within the conf file for the species. However, when i mouse
over the track, the link points back to localhost. If i change the link to
http://www.phytozome.net, the link works fine. http links on other
tracks work fine. None of the other links are https links. Any idea why
gbrowse is behaving such.
Thanks for your help.
Best
-shabari

--
------------------------------------------------------------------------
Scott Cain, Ph. D. scott at scottcain
dot net
GMOD Coordinator (http://gmod.org/) 216-392-3087
<(216)%20392-3087>
Ontario Institute for Cancer Research

Shabari Subramanian

2017-01-23 21:15:39 UTC

Permalink

Hi Scott,

Thank you for your patience on this.

I realize my mistake with the link. I had not parsed out the organism name
from the Note attribute. I made the changes on this conf file. Please find
new conf file attached.

link = sub {
my $feature = shift;
my $Name=$feature->name();
my ($NOTE) = $feature->attributes("Note");
return "https://phytozome.jgi.doe.gov/pz/portal.html#!gene?organism=
".$Note."&searchText=locusName:".$Name;
}

Below is an example of a link that works directly on the browser but does
not link to from the gbrowse track.

https://phytozome.jgi.doe.gov/pz/portal.html#!gene?organism=Malus
domestica&searchText=locusName:MDP0000293627

The link to phytozome is still being tested, so i only have it on one gene
on the test site. Here is a link to that gene being shown on our gbrowse
test site. It should be public if not please let me know.

http://back-gbrowse-test.arabidopsis.org/cgi-bin/gb2/gbrowse/arabidopsis/?name=Chr1:4,001,280..4,003,442

The track you should go to is "Eudicot Gene Families (Phytozome 11)". If
you click on it, it goes to localhost page on this locus.

Thank again for your help and your patience on this.

Best

-shabari

Post by Scott Cain
Hi Shabari,
I was a little confused by your original description of the link problem,
which was compounded by the fact that the URL you have in the config looks
like it only works for A. thaliana features, but the GFF features don't
seem to have any (assuming that's what the Note attribute means). When I
tried to manually build a function URL with the callback you provided, I
just got a not found message from Phytozome.
But, all of this is to say that I went back and read your original
description of the problem and it doesn't feel like a problem with the
GBrowse. Is it possible that you have something either in your /etc/hosts
file or at your institution that is rerouting these requests? The
substituting in localhost for the JGI domain is a little weird. Do you
have a publicly available instance of this I can look at?
Thanks,
Scott
On Mon, Jan 23, 2017 at 2:07 PM, Shabari Subramanian <

Hi Scott,
Apologies for not being clearer.
Thank you for clarifying how the initial landmark works. I am clear on
the behavior now.
Please find attached two files, the config file with the broken link to
phytozome and a second file with the gff formatted annotations file for the
gbrowse track in question.
Here is the part of the config file that i added to try to link to
phytozome.
link = sub {
my $feature = shift;
my $Name=$feature->name();
return "https://phytozome.jgi.doe.gov
/pz/portal.html#!gene?organism=Athaliana&searchText=locusName:".$Name;
}
Please let me know if you need more information.
Thanks for your help.
-shabari

Hi Scott,
I recently joined Phoenix Bioinformatics. I have been working on fixing
some bugs on the gbrowse viewer and ran into a wall. Tanya Berardini
suggested i reach out to you.
1. I set the inital landmark on the species conf file in /etc/gbrowse
to point to specific chr location. However the landmark field still comes
up blank on the gbrowse page for the species.
2. I am trying to set a link from gbrowse tracks of ortholgous genes to
the source phytozome pages spcific to the gene on the track. An example of
such a link is https://phytozome.jgi.doe.gov/
pz/portal.html#!gene?organism=Athaliana&searchText=locusName:AT1G01040.
I set this link within the conf file for the species. However, when i mouse
over the track, the link points back to localhost. If i change the link to
http://www.phytozome.net, the link works fine. http links on other
tracks work fine. None of the other links are https links. Any idea why
gbrowse is behaving such.
Thanks for your help.
Best
-shabari

--
------------------------------------------------------------------------
Scott Cain, Ph. D. scott at scottcain
dot net
GMOD Coordinator (http://gmod.org/) 216-392-3087
<(216)%20392-3087>
Ontario Institute for Cancer Research

Scott Cain

2017-01-24 16:20:41 UTC

Permalink

Hi Shabari,

This is going to sound a little condescending, but since I've done it
several times, I thought I'd suggest it: are you sure that the config file
you're editing is the one the GBrowse is actually using? Right now the
track that you're having difficulty with is not using the callback at
all--it's using the default gbrowse_details link as though there were no
"link" attribute in the track config at all. I'd suggest making some other
simple, obvious change, like the name of the organism to see if it changes
like you'd expect. Also, make sure that you're restarting apache for
changes if you're using something like fastCGI, which caches config files
in memory (preventing it from seeing changes without a restart). Also
check the apache error log for any messages that might indicate a problem
with the callback. You can also add warn statements to the callback that
would go out to the error log, though my recollection is that warn messages
don't always make it to the error log, so YMMV.

Scott

Scott

On Mon, Jan 23, 2017 at 4:15 PM, Shabari Subramanian <

Hi Scott,
Thank you for your patience on this.
I realize my mistake with the link. I had not parsed out the organism name
from the Note attribute. I made the changes on this conf file. Please find
new conf file attached.
link = sub {
my $feature = shift;
my $Name=$feature->name();
my ($NOTE) = $feature->attributes("Note");
return "https://phytozome.jgi.doe.gov/pz/portal.html#!gene?organism=
".$Note."&searchText=locusName:".$Name;
}
Below is an example of a link that works directly on the browser but does
not link to from the gbrowse track.
https://phytozome.jgi.doe.gov/pz/portal.html#!gene?organism=Malus
domestica&searchText=locusName:MDP0000293627
The link to phytozome is still being tested, so i only have it on one gene
on the test site. Here is a link to that gene being shown on our gbrowse
test site. It should be public if not please let me know.
http://back-gbrowse-test.arabidopsis.org/cgi-bin/gb2/
gbrowse/arabidopsis/?name=Chr1:4,001,280..4,003,442
The track you should go to is "Eudicot Gene Families (Phytozome 11)". If
you click on it, it goes to localhost page on this locus.
Thank again for your help and your patience on this.
Best
-shabari

Hi Scott,
Apologies for not being clearer.
Thank you for clarifying how the initial landmark works. I am clear on
the behavior now.
Please find attached two files, the config file with the broken link to
phytozome and a second file with the gff formatted annotations file for the
gbrowse track in question.
Here is the part of the config file that i added to try to link to
phytozome.
link = sub {
my $feature = shift;
my $Name=$feature->name();
return "https://phytozome.jgi.doe.gov
/pz/portal.html#!gene?organism=Athaliana&searchText=locusName:".$Name;
}
Please let me know if you need more information.
Thanks for your help.
-shabari

Hi Scott,
I recently joined Phoenix Bioinformatics. I have been working on
fixing some bugs on the gbrowse viewer and ran into a wall. Tanya Berardini
suggested i reach out to you.
1. I set the inital landmark on the species conf file in /etc/gbrowse
to point to specific chr location. However the landmark field still comes
up blank on the gbrowse page for the species.
2. I am trying to set a link from gbrowse tracks of ortholgous genes
to the source phytozome pages spcific to the gene on the track. An example
of such a link is https://phytozome.jgi.doe.gov/
pz/portal.html#!gene?organism=Athaliana&searchText=locusName:AT1G01040.
I set this link within the conf file for the species. However, when i mouse
over the track, the link points back to localhost. If i change the link to
http://www.phytozome.net, the link works fine. http links on other
tracks work fine. None of the other links are https links. Any idea why
gbrowse is behaving such.
Thanks for your help.
Best
-shabari

--
------------------------------------------------------------
------------
Scott Cain, Ph. D. scott at scottcain
dot net
GMOD Coordinator (http://gmod.org/) 216-392-3087
<(216)%20392-3087>
Ontario Institute for Cancer Research

--
------------------------------------------------------------------------
Scott Cain, Ph. D. scott at scottcain
dot net
GMOD Coordinator (http://gmod.org/) 216-392-3087
<(216)%20392-3087>
Ontario Institute for Cancer Research

Scott Cain

2017-01-24 20:00:10 UTC

Permalink

Another thing I'll do when I'm trying to debug something like this is take
the callback that is causing problems and put it in "balloon hover" or
"balloon click" so that I can see what is actually getting generated by the
callback.

Also for the truly paranoid, I'll retype the whole think including "link ="
in case the lines that are there have hidden control characters that are
wrecking things (I have a history of this actually happening to me going
back to the mid-80s, so it is somewhat justified :-) What you describe is
actually consistent with this hypothesis, though, since a stray control
character on the left hand side of the equals sign would make it look like
the "link =" was something else.

On Tue, Jan 24, 2017 at 2:50 PM, Shabari Subramanian <

Not at all. I was questioning my sanity when i first saw this behavior. :)
If you notice the other tracks on the test page i sent do go to the
expected links. Also if i modify the link to phytozome for the eudicots
track to "phytozome.net" it links out fine. I am quite sure the config
file i sent is the one gbrowse is using for the species. I did try
restarting httpd. I will check the apache error logs.

Post by Scott Cain
Hi Shabari,
This is going to sound a little condescending, but since I've done it
several times, I thought I'd suggest it: are you sure that the config file
you're editing is the one the GBrowse is actually using? Right now the
track that you're having difficulty with is not using the callback at
all--it's using the default gbrowse_details link as though there were no
"link" attribute in the track config at all. I'd suggest making some other
simple, obvious change, like the name of the organism to see if it changes
like you'd expect. Also, make sure that you're restarting apache for
changes if you're using something like fastCGI, which caches config files
in memory (preventing it from seeing changes without a restart). Also
check the apache error log for any messages that might indicate a problem
with the callback. You can also add warn statements to the callback that
would go out to the error log, though my recollection is that warn messages
don't always make it to the error log, so YMMV.
Scott
Scott
On Mon, Jan 23, 2017 at 4:15 PM, Shabari Subramanian <

Hi Scott,
Thank you for your patience on this.
I realize my mistake with the link. I had not parsed out the organism
name from the Note attribute. I made the changes on this conf file. Please
find new conf file attached.
link = sub {
my $feature = shift;
my $Name=$feature->name();
my ($NOTE) = $feature->attributes("Note");
return "https://phytozome.jgi.doe.gov/pz/portal.html#!gene?organism=
".$Note."&searchText=locusName:".$Name;
}
Below is an example of a link that works directly on the browser but
does not link to from the gbrowse track.
https://phytozome.jgi.doe.gov/pz/portal.html#!gene?organism=Malus
domestica&searchText=locusName:MDP0000293627
The link to phytozome is still being tested, so i only have it on one
gene on the test site. Here is a link to that gene being shown on our
gbrowse test site. It should be public if not please let me know.
http://back-gbrowse-test.arabidopsis.org/cgi-bin/gb2/gbrowse
/arabidopsis/?name=Chr1:4,001,280..4,003,442
The track you should go to is "Eudicot Gene Families (Phytozome 11)". If
you click on it, it goes to localhost page on this locus.
Thank again for your help and your patience on this.
Best
-shabari

Post by Scott Cain
Hi Shabari,
I was a little confused by your original description of the link
problem, which was compounded by the fact that the URL you have in the
config looks like it only works for A. thaliana features, but the GFF
features don't seem to have any (assuming that's what the Note attribute
means). When I tried to manually build a function URL with the callback
you provided, I just got a not found message from Phytozome.
But, all of this is to say that I went back and read your original
description of the problem and it doesn't feel like a problem with the
GBrowse. Is it possible that you have something either in your /etc/hosts
file or at your institution that is rerouting these requests? The
substituting in localhost for the JGI domain is a little weird. Do you
have a publicly available instance of this I can look at?
Thanks,
Scott
On Mon, Jan 23, 2017 at 2:07 PM, Shabari Subramanian <

Hi Scott,
Apologies for not being clearer.
Thank you for clarifying how the initial landmark works. I am clear on
the behavior now.
Please find attached two files, the config file with the broken link
to phytozome and a second file with the gff formatted annotations file for
the gbrowse track in question.
Here is the part of the config file that i added to try to link to
phytozome.
link = sub {
my $feature = shift;
my $Name=$feature->name();
return "https://phytozome.jgi.doe.gov
/pz/portal.html#!gene?organism=Athaliana&searchText=locusName:".$Name;
}
Please let me know if you need more information.
Thanks for your help.
-shabari

Post by Scott Cain
Hi Shabari,
I'm cc'ing the GBrowse mailing list so the discussion can be archived.
About the initial landmark: can you provide a copy of the config file
that isn't working? I think I described in an earlier email that the
initial landmark only comes into effect when someone is visiting the site
who has never been there (since after the first visit, a cookie remembers
the last cooridinates visited). If there is some other problem with the
initial landmark please describe the problem more thoroughly.
About the links out: please provide the track configuration that
isn't working and a sample of the GFF that should be affected by the
configuration. It's pretty hard to troubleshoot from your description :-)
Thanks,
Scott
On Mon, Jan 23, 2017 at 1:09 AM, Shabari Subramanian <

Hi Scott,
I recently joined Phoenix Bioinformatics. I have been working on
fixing some bugs on the gbrowse viewer and ran into a wall. Tanya Berardini
suggested i reach out to you.
1. I set the inital landmark on the species conf file in
/etc/gbrowse to point to specific chr location. However the landmark field
still comes up blank on the gbrowse page for the species.
2. I am trying to set a link from gbrowse tracks of ortholgous genes
to the source phytozome pages spcific to the gene on the track. An example
of such a link is https://phytozome.jgi.doe.gov/
pz/portal.html#!gene?organism=Athaliana&searchText=locusName
:AT1G01040. I set this link within the conf file for the species.
However, when i mouse over the track, the link points back to localhost. If
i change the link to http://www.phytozome.net, the link works fine.
http links on other tracks work fine. None of the other links are https
links. Any idea why gbrowse is behaving such.
Thanks for your help.
Best
-shabari

--
------------------------------------------------------------
------------
Scott Cain, Ph. D. scott at
scottcain dot net
GMOD Coordinator (http://gmod.org/) 216-392-3087
<(216)%20392-3087>
Ontario Institute for Cancer Research

--
------------------------------------------------------------
------------
Scott Cain, Ph. D. scott at scottcain
dot net
GMOD Coordinator (http://gmod.org/) 216-392-3087
<(216)%20392-3087>
Ontario Institute for Cancer Research

--
------------------------------------------------------------------------
Scott Cain, Ph. D. scott at scottcain
dot net
GMOD Coordinator (http://gmod.org/) 216-392-3087
<(216)%20392-3087>
Ontario Institute for Cancer Research

Shabari Subramanian

2017-01-24 20:11:17 UTC

Permalink

Thanks for your help. I will keep at it.

best

-shabari

Post by Scott Cain
Another thing I'll do when I'm trying to debug something like this is take
the callback that is causing problems and put it in "balloon hover" or
"balloon click" so that I can see what is actually getting generated by the
callback.
Also for the truly paranoid, I'll retype the whole think including "link
=" in case the lines that are there have hidden control characters that are
wrecking things (I have a history of this actually happening to me going
back to the mid-80s, so it is somewhat justified :-) What you describe is
actually consistent with this hypothesis, though, since a stray control
character on the left hand side of the equals sign would make it look like
the "link =" was something else.
On Tue, Jan 24, 2017 at 2:50 PM, Shabari Subramanian <

Hi Scott,
Thank you for your patience on this.
I realize my mistake with the link. I had not parsed out the organism
name from the Note attribute. I made the changes on this conf file. Please
find new conf file attached.
link = sub {
my $feature = shift;
my $Name=$feature->name();
my ($NOTE) = $feature->attributes("Note");
return "https://phytozome.jgi.doe.gov/pz/portal.html#!gene?organism=
".$Note."&searchText=locusName:".$Name;
}
Below is an example of a link that works directly on the browser but
does not link to from the gbrowse track.
https://phytozome.jgi.doe.gov/pz/portal.html#!gene?organism=Malus
domestica&searchText=locusName:MDP0000293627
The link to phytozome is still being tested, so i only have it on one
gene on the test site. Here is a link to that gene being shown on our
gbrowse test site. It should be public if not please let me know.
http://back-gbrowse-test.arabidopsis.org/cgi-bin/gb2/gbrowse
/arabidopsis/?name=Chr1:4,001,280..4,003,442
The track you should go to is "Eudicot Gene Families (Phytozome 11)".
If you click on it, it goes to localhost page on this locus.
Thank again for your help and your patience on this.
Best
-shabari

Post by Scott Cain
Hi Shabari,
I was a little confused by your original description of the link
problem, which was compounded by the fact that the URL you have in the
config looks like it only works for A. thaliana features, but the GFF
features don't seem to have any (assuming that's what the Note attribute
means). When I tried to manually build a function URL with the callback
you provided, I just got a not found message from Phytozome.
But, all of this is to say that I went back and read your original
description of the problem and it doesn't feel like a problem with the
GBrowse. Is it possible that you have something either in your /etc/hosts
file or at your institution that is rerouting these requests? The
substituting in localhost for the JGI domain is a little weird. Do you
have a publicly available instance of this I can look at?
Thanks,
Scott
On Mon, Jan 23, 2017 at 2:07 PM, Shabari Subramanian <

Hi Scott,
Apologies for not being clearer.
Thank you for clarifying how the initial landmark works. I am clear
on the behavior now.
Please find attached two files, the config file with the broken link
to phytozome and a second file with the gff formatted annotations file for
the gbrowse track in question.
Here is the part of the config file that i added to try to link to
phytozome.
link = sub {
my $feature = shift;
my $Name=$feature->name();
return "https://phytozome.jgi.doe.gov
/pz/portal.html#!gene?organism=Athaliana&searchText=locusName
:".$Name;
}
Please let me know if you need more information.
Thanks for your help.
-shabari

Post by Scott Cain
Hi Shabari,
I'm cc'ing the GBrowse mailing list so the discussion can be archived.
About the initial landmark: can you provide a copy of the config
file that isn't working? I think I described in an earlier email that the
initial landmark only comes into effect when someone is visiting the site
who has never been there (since after the first visit, a cookie remembers
the last cooridinates visited). If there is some other problem with the
initial landmark please describe the problem more thoroughly.
About the links out: please provide the track configuration that
isn't working and a sample of the GFF that should be affected by the
configuration. It's pretty hard to troubleshoot from your description :-)
Thanks,
Scott
On Mon, Jan 23, 2017 at 1:09 AM, Shabari Subramanian <

Hi Scott,
I recently joined Phoenix Bioinformatics. I have been working on
fixing some bugs on the gbrowse viewer and ran into a wall. Tanya Berardini
suggested i reach out to you.
1. I set the inital landmark on the species conf file in
/etc/gbrowse to point to specific chr location. However the landmark field
still comes up blank on the gbrowse page for the species.
2. I am trying to set a link from gbrowse tracks of ortholgous
genes to the source phytozome pages spcific to the gene on the track. An
example of such a link is https://phytozome.jgi.doe.gov/
pz/portal.html#!gene?organism=Athaliana&searchText=locusName
:AT1G01040. I set this link within the conf file for the species.
However, when i mouse over the track, the link points back to localhost. If
i change the link to http://www.phytozome.net, the link works
fine. http links on other tracks work fine. None of the other links are
https links. Any idea why gbrowse is behaving such.
Thanks for your help.
Best
-shabari

--
------------------------------------------------------------
------------
Scott Cain, Ph. D. scott at
scottcain dot net
GMOD Coordinator (http://gmod.org/) 216-392-3087
<(216)%20392-3087>
Ontario Institute for Cancer Research

--
------------------------------------------------------------------------
Scott Cain, Ph. D. scott at scottcain
dot net
GMOD Coordinator (http://gmod.org/) 216-392-3087
<(216)%20392-3087>
Ontario Institute for Cancer Research