Closed smoe closed 2 years ago
@smoe my best guess is a namespace clash issue. group_by function from some other library might be interfering. We will look into this. @dohmjan We should make an explicit call here including the library name where group_by
function comes from.
Hey! I added the explicit calls for plotly functions. There might have been mix-ups with other libraries as I couldn't reproduce the error using guix. Hope this resolves the problem! https://github.com/BIMSBbioinfo/pigx_rnaseq/commit/b1346273d3b3756d0532fa2e7e07eaaef644cc22
Thank you tons! I also thought that this is not perfectly reproducible. Please expect feedback next week.
@smoe do you have any issues with this still?
Sorry, let me revisit your new release (congrats, btw).
@smoe thank you! Just a warning: the default mapper in the new release is set to "hisat2", which consumes much less memory. This can be configured in the settings.yaml file if you'd like to keep using STAR.
Hisat2 just ran fine, and so does STAR, but in either configuration right afterwards "samtools index" is somehow having an incompatibility with snakemake, reported as https://github.com/BIMSBbioinfo/pigx_rnaseq/issues/96. I was a bit surprised by your comment on STAR's memory footprint - had considered this an issue from the past.
Please leave this bug open for another while, since IIRC the problem was not triggered by the regular build tests.
@smoe shall I close this issue? Is it fixed for you?
My memory is somewhat weak on this - I have escapted to a machine with more memory for my typical pigx-rnaseq runs and there everything completes successfully
Hello again, I get this error message for the preparation of section 5.3 of the salmon and the star reports:
Does this ring any bell on your end? The corresponding code is in https://github.com/BIMSBbioinfo/pigx_rnaseq/blob/1375b63be6d39a58388d8bf2adc62997645d81fa/scripts/deseqReport.Rmd.in#L421 in the section generating the boxplots. The volcano plot in 5.1 and the others in 4 are fine.
I am running a self-compiled version on Debian. There are likely differences with the JavaScript and R libraries that I am using. The files generated during the autotesting did not create any R error, though.
My next step would be to prepare a guix installation and compare with the same data set. But you may have an idea already. I found that the quality of the sequencing of this example (picked E-GEOD-49555) is bad (<30 mapped for all except two samples) which may cause a confusion when grouping when one group does not have the gene?!? Just a speculation.
Thanks Steffen