Hi,
The latest version of the MSI combine tool (Galaxy version 3.4.3+galaxy0) fails depending on the order that the two imzml ("control" and "treat1") are selected as input datasets in the GTN tutorial: Mass spectrometry imaging: Finding differential analytes. If the two input imzml datasets are selected in the opposite order "treat1" and "control", then the tool runs successfully. The tool also runs successfully if an earlier version is used (e.g. Galaxy version 2.10.0.0) regardless of which order the two datasets are selected.
I am able to produce the same results (run successfully vs job failing) on both Galaxy Australia and Galaxy EU, so this is not a Galaxy server specific issue. I have included below links to my Galaxy Histories on Galaxy Australia and Galaxy EU that produce this error.
I think the issue has to do with the latest version of MSI combine (3.4.3+galaxy0) not being able to combine datasets when the second dataset is larger than the first one. The tool puts all input datasets on a single grid and in the tutorial the "control" dataset is spatially smaller than the "treat1" dataset. So when using tool version 3.4.3+galaxy0, MSI combine is failing if it first puts the smaller "control" dataset onto a grid and then tries to add the larger "treat1" dataset. I am assuming that the height of the combined dataset grid is being defined by the first dataset and is not being increased when adding a larger dataset. In contrast, when starting with the larger "treat1" dataset, there is no additional height required as the "control" dataset is 1/3 the height.
Additionally, when MSI combine fails, it provides the following error message:
Error in order(y) : unimplemented type 'list' in 'orderVector1'
Calls: do.call ... cbind2 -> atoms -> as.factor -> as.factor -> factor -> order
Execution halted
Thanks,
Tristan
Hi,
The latest version of the
MSI combinetool (Galaxy version3.4.3+galaxy0) fails depending on the order that the two imzml ("control" and "treat1") are selected as input datasets in the GTN tutorial: Mass spectrometry imaging: Finding differential analytes. If the two input imzml datasets are selected in the opposite order "treat1" and "control", then the tool runs successfully. The tool also runs successfully if an earlier version is used (e.g. Galaxy version2.10.0.0) regardless of which order the two datasets are selected.I am able to produce the same results (run successfully vs job failing) on both Galaxy Australia and Galaxy EU, so this is not a Galaxy server specific issue. I have included below links to my Galaxy Histories on Galaxy Australia and Galaxy EU that produce this error.
I think the issue has to do with the latest version of MSI combine (
3.4.3+galaxy0) not being able to combine datasets when the second dataset is larger than the first one. The tool puts all input datasets on a single grid and in the tutorial the "control" dataset is spatially smaller than the "treat1" dataset. So when using tool version3.4.3+galaxy0, MSI combine is failing if it first puts the smaller "control" dataset onto a grid and then tries to add the larger "treat1" dataset. I am assuming that the height of the combined dataset grid is being defined by the first dataset and is not being increased when adding a larger dataset. In contrast, when starting with the larger "treat1" dataset, there is no additional height required as the "control" dataset is 1/3 the height.Additionally, when
MSI combinefails, it provides the following error message:Thanks,
Tristan