amp_boxplots - method to determine significance for taxa between designated groups?

[reige012]

Hi, I'm wondering if there is a way to get the significance for taxa between designated groups when running the amp_boxplot() command. I see that there is an "extra" feature called amp_diffabund, but its not available for R 3.5 versions. Is there any other way to access this data or at least the accurate average values so that I can run significance tests on my own?

Thank you, Alicia

[KasperSkytte]

Hi Alicia

The boxplot is just a visualisation of the data, there is no fancy statistics going on other than quartiles+median. But if you want to test for differences between groups of samples (differential abundance) you can use the amp_diffabund() function. It's only available from the ampvis2extras packages as it is no longer being maintained, at least actively, however. Mainly because it's based on the DESeq2 package which is developed for transcriptomics, and is not really ideal for amplicon data since it doesn't deal well with compositional data compared to fx ANCOM or ANCOM2 fx. There is some sense to it though, major conclusions would probably still make sense.

I just updated the package a bit, installation instructions were outdated and amp_mergereplicates() has been included in ampvis2 for a while now, so was also outdated. See installation instruction at https://kasperskytte.github.io/ampvis2extras/

[reige012]

Hi Kasper,

Thank you for the information. Would be awesome if there was an option in the box plot to at least extract those data.

Amp_diffabund() was my first thought since I've already got the data into the correct format for ampvis2.

I did just now try to reinstall it and got an error. I used the install instructions at the above link.

Bioconductor version 3.8 (BiocManager 1.30.10), R 3.5.2 (2018-12-20) Installing github package(s) 'kasperskytte/ampvis2extras' Error: Failed to install 'unknown package' from GitHub: JSON: EXPECTED value GOT <

Any chance you know why I got the error?

Thank you again for your help. Alicia

On Thu, Apr 23, 2020 at 6:19 AM Kasper Skytte Andersen < notifications@github.com> wrote:

Hi Alicia

The boxplot is just a visualisation of the data, there is no fancy statistics going on other than quartiles+median. But if you want to test for differences between groups of samples (differential abundance) you can use the amp_diffabund() function. It's only available from the ampvis2extras packages as it is no longer being maintained, at least actively, however. Mainly because it's based on the DESeq2 package which is developed for transcriptomics, and is not really ideal for amplicon data since it doesn't deal well with compositional data compared to fx ANCOM or ANCOM2 fx. There is some sense to it though, major conclusions would probably still make sense.

I just updated the package a bit, installation instructions were outdated and amp_mergereplicates() has been included in ampvis2 for a while now, so was also outdated. See installation instruction at https://kasperskytte.github.io/ampvis2extras/

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/MadsAlbertsen/ampvis2/issues/89#issuecomment-618343795, or unsubscribe https://github.com/notifications/unsubscribe-auth/AD7BP7ZQO46NPBAQMARBT7DROAP4BANCNFSM4MOL5VEA .

-- Ph.D. Candidate, Louisiana State University | Michael E.Hellberg Lab M.S. Biology, Georgia Southern University | Daniel F. Gleason Lab Lab Phone: (225) 578-9114 Email: areige1@lsu.edu Website: aliciamreigel.wordpress.com

[KasperSkytte]

Hmm that is a funny error I have never seen. I can install without problems. I have a different R version though. After the error, what does traceback() return?

[reige012]

Whoops, I got this error resolved by calling devtools before running the installation.

Thanks again for your help, its working now!

On Thu, Apr 23, 2020 at 8:41 AM Alicia Reigel reige012@gmail.com wrote:

Hi Kasper,

Thank you for the information. Would be awesome if there was an option in the box plot to at least extract those data.

Amp_diffabund() was my first thought since I've already got the data into the correct format for ampvis2.

I did just now try to reinstall it and got an error. I used the install instructions at the above link.

Bioconductor version 3.8 (BiocManager 1.30.10), R 3.5.2 (2018-12-20) Installing github package(s) 'kasperskytte/ampvis2extras' Error: Failed to install 'unknown package' from GitHub: JSON: EXPECTED value GOT <

Any chance you know why I got the error?

Thank you again for your help. Alicia

On Thu, Apr 23, 2020 at 6:19 AM Kasper Skytte Andersen < notifications@github.com> wrote:

Hi Alicia

The boxplot is just a visualisation of the data, there is no fancy statistics going on other than quartiles+median. But if you want to test for differences between groups of samples (differential abundance) you can use the amp_diffabund() function. It's only available from the ampvis2extras packages as it is no longer being maintained, at least actively, however. Mainly because it's based on the DESeq2 package which is developed for transcriptomics, and is not really ideal for amplicon data since it doesn't deal well with compositional data compared to fx ANCOM or ANCOM2 fx. There is some sense to it though, major conclusions would probably still make sense.

I just updated the package a bit, installation instructions were outdated and amp_mergereplicates() has been included in ampvis2 for a while now, so was also outdated. See installation instruction at https://kasperskytte.github.io/ampvis2extras/

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/MadsAlbertsen/ampvis2/issues/89#issuecomment-618343795, or unsubscribe https://github.com/notifications/unsubscribe-auth/AD7BP7ZQO46NPBAQMARBT7DROAP4BANCNFSM4MOL5VEA .

-- Ph.D. Candidate, Louisiana State University | Michael E.Hellberg Lab M.S. Biology, Georgia Southern University | Daniel F. Gleason Lab Lab Phone: (225) 578-9114 Email: areige1@lsu.edu Website: aliciamreigel.wordpress.com

-- Ph.D. Candidate, Louisiana State University | Michael E.Hellberg Lab M.S. Biology, Georgia Southern University | Daniel F. Gleason Lab Lab Phone: (225) 578-9114 Email: areige1@lsu.edu Website: aliciamreigel.wordpress.com

[KasperSkytte]

Good, you're welcome