Test failing

[pkrog]

The test test_basic_processing() is failing inside function chooseVerticesLosses(), in line:

    vmz <- lf@formula[bloss,] %*% m_atoms

where m_atoms is an empty numeric vector.

--> This one is solved. atoms was a integer vector with atom letters as names instead of a character vector.

[pkrog]

Now we have:

object 'bf' not found
1: test_basic_processing()
2: mineMS2::plotPatterns(m2l, full = TRUE) at /home/pr228844/dev/mineMS2_pkg/tests/testthat/test.R:36
3: plot(m2l, id) at /home/pr228844/dev/mineMS2_pkg/R/complexPlot.R:68
4: plot(m2l, id)
5: .local(x, y, ...)
6: plot(x[y], title = y, dags = mm2Dags(x), edgeLabels = (mm2EdgesLabels(x)), atoms = names(x@atoms),
       formula = get_formula(x)[toccs], ...) at /home/pr228844/dev/mineMS2_pkg/R/ms2Lib.R:608
7: plot(x[y], title = y, dags = mm2Dags(x), edgeLabels = (mm2EdgesLabels(x)), atoms = names(x@atoms),
       formula = get_formula(x)[toccs], ...)
8: .local(x, y, ...)
9: makeLabelPattern(x, atoms, dags, edgeLabels, oformula = oformula) at /home/pr228844/dev/mineMS2_pkg/R/fragPattern-plot.R:225
10: annotateAFG(p, atoms, dags, elabs, oformula = oformula) at /home/pr228844/dev/mineMS2_pkg/R/labels.R:531
11: annotateVertices(fp, label_origin, dags, elabs, atoms, allf = allf[pmm], edge_label = "lab",
       oformula = oformula, massdiff = massdiff[pmm]) at /home/pr228844/dev/mineMS2_pkg/R/labels.R:508
12: sapply(seq_along(vlab), function(x, l_origin, voformula, vatoms, subf, vdags, lformula,
       vmz, vrdbe) {
       lf <- lformula[[x]]
       if (length(lf) == 0) {
           return(list(c(1, 1), lf))
       }
       vrdbe <- vecRDBE(lf@formula)
       rtemp <- chooseVerticesLosses(lf, oformula = voformula, dags, vmz[x], vrdbe = vrdbe,
           atoms = vatoms, subformula = subf[x + 1, ])
       return(list(rtemp[2:3], lf[rtemp[1], , drop = FALSE]))
   }, vdags = vdags, voformula = oformula, l_origin = vlab, vatoms = atoms, subf = subformula,
       lformula = allf, vmz = mzv, vrdbe = vrdbe) at /home/pr228844/dev/mineMS2_pkg/R/labels.R:282
13: lapply(X = X, FUN = FUN, ...)
14: FUN(X[[i]], ...)
15: chooseVerticesLosses(lf, oformula = voformula, dags, vmz[x], vrdbe = vrdbe, atoms = vatoms,
       subformula = subf[x + 1, ]) at /home/pr228844/dev/mineMS2_pkg/R/labels.R:290

It happens in function chooseVerticesLosses() at the following line:

bpartial <- as.numeric(as.numeric(bf[bloss])==nrow(lf@formula))

@adelabriere , could you look at that please or explain me what to do?

[pkrog]

What should bf contain?

[adelabriere]

Hi Pierrick, thanks for reporting, bf is a vector containing the number of vertices/subformula matched for each candidates formmula (oformula). Is this still an issue ?

[pkrog]

Hi Alexis, I've solved, I think. Please look at:
https://github.com/adelabriere/mineMS2/pull/8/files#diff-d8566cf7b3fb7f63dfb157edc80e142cR236
and tell me if it looks right or wrong.

The output of my test is different from the output of the reference file of the Galaxy tool. The changed lines are the v_id lines:

<data key="v_id">P15,P15</data>  

In the case of the line above, P15,P15 is replaced with P27,P27. Is it important?

[adelabriere]

This does not seems to be an issue. I checked the results on the vignette and this does not have any practical consequence.

I probably changed a default parameters at some point for clarity purpose.

[pkrog]

The IDs in the output are different on each run. Do you use some random numbers in your algo? Can we set a seed for testing?

[adelabriere]

Just a quick update I m investigating this behavior at the moment. I don t see any evident reason in the R code, so it is probably coming from the C++ code.

[adelabriere]

Issu solved, pushed on the master branch, I had the same ID on 10 tests so if you don t encounter this issue anymore I ll consider that the problem is solved.

[pkrog]

Ok, I'll look at that. Thanks.

[pkrog]

Maybe I was in a bad luck, but I obtained this difference between two runs:

678c678                                                                                                                                                                                        
<       <data key="v_id">P168</data>                                                                                                                                                           
---                                                                                                                                                                                            
>       <data key="v_id">P166</data>                                                                                                                                                           
694c694                                                                                                                                                                                        
<       <data key="v_id">P168</data>                                                                                                                                                           
---                                                                                                                                                                                            
>       <data key="v_id">P166</data>                                                                                                                                                           

I just ran make test to run this test. The files to compare are:
tests/output/test_basic_processing.graphml
and
tests/res/basic_annotated.graphml

See branch basic_test.

[adelabriere]

OK I check that, It s clearly not luck as the IDs are now close.

[pkrog]

I add @ethevenot to this conversation.

[pkrog]

Hi @adelabriere , did you have time to look at this last small issue ?