Question

GOexpress - Trouble with subset_scores result (<0 rows>)

0

Entering edit mode

sina.zboron • 0

@sinazboron-14859

Last seen 6.9 years ago

Hi,

I was applying the "GOexpress: Visualise gene expression data using gene ontology annotations" (from October 30, 2017) step by step to my microarray data and everything seems to be working fine, except when I tried to filter my result with the subset_scores function.

I checked the results from the previous two steps (random forest algorithm using local annotations, permutation-based P-value for ontologies) and they nicely compare to the example in their structure and content, but when I go on with filtering I receive a list as a result containing the relevant columnnames but 0 rows. ("<0 rows> (or 0-length row.names)")

I manually checked whether my input list even contains rows that fit to the filter criteria, and yes, there should be several rows that match (total_count=5, p.val=0.05).

Thanks in advance!

Regards, Sina

GOexpress subset_scores <0 rows> • 1.5k views

ADD COMMENT • link 6.9 years ago sina.zboron • 0

0

Entering edit mode

Dear Sina,

Any chance you can email me a minimal reproducible example (MWE) of the issue, so that I can examine the issue myself? If possible the input list that you are working with?
You can export the list as follows:
```
saveRDS(name_of_your_list, "file_name.rds")
```
and email me the file.

My email is available if you enter at the R prompt:
```
maintainer("GOexpress")
```

Otherwise, any example that I can use to reproduce the error will do.

Thank you,
Kevin

ADD REPLY • link 6.9 years ago kevin.rue ▴ 350

score 1 · Accepted Answer · 2018-01-25

Dear Sina.

Having looked at the objects that you sent me, I can say that this is due to some hard-coded assumptions in the code of GOexpress. It expects values in the column namespace_1003 to be one of c("biological_process","molecular_function","cellular_component"). To clarify, BP, MF, and CC are only widely used abbreviations that GOexpress supports, but translates within the function into their longer name.

You will have to run the code below before you use subset_scores on your data set:

ExpSetAB_result.pVal$GO$namespace_1003 <- factor(

  ExpSetAB_result.pVal$GO$namespace_1003,
  levels = c("BP","MF","CC"),
  labels = c("biological_process","molecular_function","cellular_component")
)

Can you please tell me where you obtained your annotations from? I can see that they include an ontology "universal" that I have never seen before (at least not in Gene Ontology).

> levels(gores$GO$namespace_1003)
[1] "BP"        "CC"        "MF"        "universal"

Best wishes,
Kevin

score 1 · Accepted Answer · 2018-01-25

1

Entering edit mode

sina.zboron • 0

@sinazboron-14859

Last seen 6.9 years ago

Dear Kevin,

thank you so much for your help - it works! Initially I tried it with "biological_function" as described, but it didn't work :-)

For the GO (see http://manuals.bioinformatics.ucr.edu/home/R_BioCondManual#TOC-Gene-Ontologies):

go_df <- data.frame(GOID=unlist(eapply(GOTERM, function(x) x@GOID)), Term=unlist(eapply(GOTERM, function(x) x@Term)), Ont=unlist(eapply(GOTERM, function(x) x@Ontology)))

Kind regards,

Sina

ADD COMMENT • link 6.9 years ago sina.zboron • 0

0

Entering edit mode

Thank you for your feedback. I'm glad I could solve your issue and I'll see what I can do to address it in the future. The resource that you linked will be useful for this.
Can I ask you to click the button that "accepts" my answer above, so that future visitor to the post can identify the solution immediately?
Many thanks
Kevin

ADD REPLY • link 6.9 years ago kevin.rue ▴ 350