DA with IBD dataset using mbzinb.test

[xix017]

Hello Jun, I used the example of mbznb.test to do differential analysis on the IBD dataset and only 5 OTUs were found to be significant (Padj<0.05), which was much less than the number described in the paper (48 significant OTUs were found with omnibus test). Could you please tell me is there any parameter should be tuned to get the same output as the paper? Below is the code:

library(mbzinb) library(dplyr) data(IBDdat) mbzinb.test(IBDdat, group="ULCERATIVE_COLIT_OR_CROHNS_DIS") %>% mbzinb.results() %>% arrange(Padj) %>% View()

Thanks a lot for your time and help. Best, Min

[jchen1981]

We used the GMPR normalization in 'mbzinb.test' instead of the default RLE. The GMPR size factor could be calculated using https://github.com/jchen1981/GMPR

Also see the reference https://peerj.com/articles/4600/

We leave the normalization method open in 'mbzinb.test' since it is crucial for addressing compositional effects and new methods are being developed.

Best, Jun

On Tue, Feb 18, 2020 at 10:11 PM xiaomin xu notifications@github.com wrote:

Hello Jun, I used the example of mbznb.test to do differential analysis on the IBD dataset and only 5 OTUs were found to be significant (Padj<0.05), which was much less than the number described in the paper (48 significant OTUs were found with omnibus test). Could you please tell me is there any parameter should be tuned to get the same output as the paper? Below is the code:

library(mbzinb) library(dplyr) data(IBDdat) mbzinb.test(IBDdat, group="ULCERATIVE_COLIT_OR_CROHNS_DIS") %>% mbzinb.results() %>% arrange(Padj) %>% View()

Thanks a lot for your time and help. Best, Min

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/jchen1981/MicrobiomeDDA/issues/3?email_source=notifications&email_token=ABHOPVS22VMF2M2OSYUAKRTRDSWONA5CNFSM4KXRMESKYY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4IOQLOPQ, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABHOPVW3HYIPSNYCIPU267TRDSWONANCNFSM4KXRMESA .

[xix017]

Hi Jun, Thanks for your quick answer. However, I could find the parameter to set the normalization method in mbzinb.test. Below is the parameter given： mbzinb.test(mbzinb.data, group, filter.if.unfiltered = TRUE, LRT = "omnibus", outlier.option = "winsor", upper.p = 0.97, cooks.cutoff = 1, use.zeroinfl = FALSE)

This parameter is given in ZISeq. However, even though I used ZISeq with GMPR normalization to do DA on the IBD dataset, there were only 11 OTUs to be found significantly different. Also, while running ZISeq, I have to remove sample 100139.518500 to enable it to work. Is the IBD dataset loaded from mbzinb library same as the one you used in the paper An omnibus test for differential distribution analysis of microbiome sequencing data?

Thanks again for your time and help. Best, Min

[jchen1981]

Hi Min,

Please try to use mbzinb::norm.factors(mbzinb.data, norm.factors = gmpr.norm.factor) to set the normalization factor. The IBD dataset included in the package only contained a subset of OTUs for demonstration purposes. In the manuscript, we kept all OTUs with prevalence > 10% and removed one sample that did not have sufficient number of reads (see Table 2). We also used 10% FDR.

We will update the mbzinb package to reflect this practice in the near future. Let me know if you have any questions.

On Thu, Feb 20, 2020 at 4:39 AM xiaomin xu notifications@github.com wrote:

Hi Jun, Thanks for your quick answer. However, I could find the parameter to set the normalization method in mbzinb.test. Below is the parameter given： mbzinb.test(mbzinb.data, group, filter.if.unfiltered = TRUE, LRT = "omnibus", outlier.option = "winsor", upper.p = 0.97, cooks.cutoff = 1, use.zeroinfl = FALSE)

This parameter is given in ZISeq. However, even though I used ZISeq with GMPR normalization to do DA on the IBD dataset, there were only 11 OTUs to be found significantly different. Also, while running ZISeq, I have to remove sample 100139.518500 to enable it to work. Is the IBD dataset loaded from mbzinb library same as the one you used in the paper An omnibus test for differential distribution analysis of microbiome sequencing data?

Thanks again for your time and help. Best, Min

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/jchen1981/MicrobiomeDDA/issues/3?email_source=notifications&email_token=ABHOPVXU6I6HMKM6PBEJXYTRDZMVZA5CNFSM4KXRMESKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEMNGU3A#issuecomment-588933740, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABHOPVWSAWGZVATHSKMUMD3RDZMVZANCNFSM4KXRMESA .