report annotations to obsoleted FYPO terms in logs

[pombase-admin]

Hi Kim,

Midori will be obsoleteing some FYPO population terms.

Can any obsoleted terms used in annotation be reported in a log file?

Many thanks

Val

Original comment by: ValWood

[pombase-admin]

These will have replaced_by tags, so the annotations can automatically be remapped to the new term.

However, we would also like a warning that the remapping has taken place, so that the remapped terms can be checked to ensure that they have the correct "cellular phenotype annotation + penetrance" (the warning should comprise of the session ID gene/ old FYPO ID-term and the new FYPO ID-term)

This is quite high priority as the terms were obsoleted yesterday. I am in the process of remapping the data from Jacky's data in pombe-embl/phenotype_mapping/one_line_per_annotation I am about 2/3 of the way through and I think the rest will take me about 4 hours so I hope to finish this tonight. I'll add a comment to this ticket when I am done. Most of the rest of the obsoleted term should be in Canto sessions.

Original comment by: ValWood

[pombase-admin]

These will have replaced_by tags

I don't think we have mentioned these before. We can re-map FYPO annotation directly if an annotation is obsoleted and there is a replaced by tag. We should Never use these automatically for GO because GOC has a habit of obsoleting a specific term and suggesting a broader term which is correct, but we lose annotation specificity. For GO we would always check as usually a more specific annotation can be made than the one that is suggested, so for GO its a warning only.

That's probably a pain....

Original comment by: ValWood

[pombase-admin]

I finished remapping Jacky's data so none of the obsolete terms should remain. It took a while as I needed to check that the penetrance was specified on the cellular phenotype annotation where possible. There is a new file pombe-embl/phenotype_mapping/one_line_per_annotation committed to svn ready to regenerate phaf_format_phenotypes.tsv Please let me know the lines which don't make it through to the phaf, and then I'll fix the things which get through to the logs after the next update.

Cheers

Val

Original comment by: ValWood

[pombase-admin]

The current load code explicitly ignores terms that are obsolete, so if you use one it will say something like: can't load annotation, FYPO:0002195 not found in database at line 2

when loading a PHAF file.

I'll change it so that says the term is obsolete rather than not found in that case.

Original comment by: kimrutherford

[pombase-admin]

That's probably a pain....

It a bit awkward at the moment. The GMOD OBO Chado loader stores the obsolete flag, but doesn't store the replaced_by tag in Chado. So when when get to loading the annotation there's no easy way to do this without loading the OBO again.

I'll put it on the list of things to add to the OBO loader I've half written.

I'll change it so that says the term is obsolete rather than not found in that case.

That's done so at least you get a sensible warning now.

Original comment by: kimrutherford

[pombase-admin]

So, just to confirm, we now have a message in the logs which tell us if a term is obsolete? and we know which canto session or phaf file

I can see the errors from my PHAF file http://curation.pombase.org/dumps/pombase-build-2013-11-26-v4-l1/logs /log.2013-11-29-02-54-08.phenotypes_from_PMID_23697806 ...I'll fix these

but I can't see any from the canto sessions (I'm sure there will be some and Antonia can begin to look at these)

Original comment by: ValWood

[pombase-admin]

There has always been a warning, it just said "term not found" instead of "term obsolete".

I suspect the sessions warning still say term not found. I'll fix it to say term obsolete instead.

Original comment by: kimrutherford

[pombase-admin]

I suspect the sessions warning still say term not found. I'll fix it to say term obsolete instead.

Fixed for the next load.

Original comment by: kimrutherford

[pombase-admin]

I see them now, I was looking in the wrong log file. It's this one http://curation.pombase.org/dumps/pombase-build-2013-11-26-v4-l1/logs/log.2013-11-29-02-54-08.curation-tool-data-load-output

(Antonia there are not so many....)

v

Original comment by: ValWood

[pombase-admin]

• Description has changed:

Diff:

--- old
+++ new
@@ -1,4 +1,3 @@
-

 Hi Kim,

• Priority: 8 --> 9

Original comment by: ValWood

[pombase-admin]

maybe isn't done yet. This is related to the item on the canto tracker upped by Antonia

Original comment by: ValWood

[pombase-admin]

Reporting at all is done but the more helpful wording isn't:

e.g. http://curation.pombase.org/dumps/pombase-build-2013-12-13-v2-l1/logs/log.2013-12-17-02-54-06.curation-tool-data-load-output

has "can't load annotation, FYPO:0002075 not found in database" and a bunch more lines like it. It doesn't (yet) say "FYPO:0002075 is obsolete".

Original comment by: mah11

[pombase-admin]

Reporting at all is done but the more helpful wording isn't:

Sorry, I thought I'd done that. I'll fix it.

Original comment by: kimrutherford

[pombase-admin]

• Description has changed:

Diff:

--- old
+++ new
@@ -1,4 +1,3 @@
-
 Hi Kim,

 Midori will be obsoleteing some FYPO  population terms.

Original comment by: kimrutherford

[pombase-admin]

You should now see messages like:

warning in ade4e74196f652e5: can't load annotation, FYPO:0002075 is an obsolete term

If it looks OK, can you close this ticket?

Original comment by: kimrutherford

[pombase-admin]

• labels: --> logs
• status: open --> closed-fixed
• assigned_to: Kim Rutherford

Original comment by: mah11

[pombase-admin]

looks fine to me

Original comment by: mah11