Random Effects / Repeated Measures

[Freemacl]

I am hoping to use ANCOMBC2 to analyse differences in my microbial community, however based on the output I'm getting, I am not sure I am properly using random effects. I have two groups of animals, vaccinated (n=19) and unvaccinated (n=8), that I sampled at 5 different time points. When I call the ANCOMBC program without the use of random effects, I do get the expected output of differentially abundant organisms that is logical enough (i.e. is consistent with other results in the project). However, when I use a random term to represent the animal ID, the output I get is exactly the same for all taxa listed (LFC = 0, se = 0.5, W= 0, p=1, diff= FALSE). Can you help me identify the problem or interpret this result?

In the code below :

• Group= timepoint
• Status= vacc/unvacc
• Rep = animal ID

ancombc2(data = physeq, 
                   tax_level = NULL,
                   fix_formula = "Group + Status",
                   rand_formula = "(1|Rep)",
                   p_adj_method = "holm", 
                   pseudo_sens = TRUE,
                   prv_cut = 0.10, 
                   lib_cut = 1000, 
                   s0_perc = 0.05,
                   group = "Group",
                   struc_zero = FALSE, 
                   neg_lb = FALSE,
                   alpha = 0.05,
                   n_cl = 2,
                   verbose = TRUE,
                   global = FALSE, 
                   pairwise = TRUE,
                   trend = FALSE,
                   iter_control = list(tol = 1e-5, max_iter = 20, verbose = FALSE),
                   em_control = list(tol = 1e-5, max_iter = 100),
                   lme_control = NULL, 
                   mdfdr_control = list(fwer_ctrl_method = "holm", B = 100), 
                   trend_control = NULL)

[FrederickHuangLin]

Hello @Freemacl,

Thank you for reaching out! The unusual result you've encountered is indeed puzzling. To gain a better understanding of the issue, I recommend running your microbial abundance data through the linear mixed effects model using lmerTest (https://cran.r-project.org/web/packages/lmerTest/index.html). Please utilize the same fixed and random formula. Could you then tell me what the outcomes are?

Best regards,
Huang

[Freemacl]

Hi Huang, thank you for getting back to me. I can definitely try that out but I'm not sure what response variable would be the most logical choice for that test. Would the abundance of one taxa work?

[FrederickHuangLin]

Yes, you can try multiple taxa, one by one, and see how the results are

[Freemacl]

Ok, I tried what you suggested, using the same formula and the lmertest package:
lmer(ASV1 ~ Treatment + Status + (1 | Rep), test)

Which produces the following output:

REML criterion at convergence: 2776.8

Scaled residuals: 
    Min      1Q  Median      3Q     Max 
-1.7836 -0.5627 -0.1532  0.2552  3.1181 

Random effects:
 Groups   Name        Variance  Std.Dev.
 Rep      (Intercept)  11436054  3382   
 Residual             102777826 10138   
Number of obs: 135, groups:  Rep, 27

Fixed effects:
                        Estimate Std. Error       df t value Pr(>|t|)    
(Intercept)             10102.20    2654.10    68.08   3.806 0.000305 ***
TreatmentPost-Challenge  3955.56    2759.20   104.00   1.434 0.154689    
TreatmentPost Vacc      -9102.30    2759.20   104.00  -3.299 0.001330 ** 
TreatmentPre-Challenge   8926.37    2759.20   104.00   3.235 0.001631 ** 
TreatmentPre Vacc       -8267.11    2759.20   104.00  -2.996 0.003418 ** 
StatusVacc                144.62    2383.85    25.00   0.061 0.952108    
---
Correlation of Fixed Effects:
            (Intr) TrtmntPs-C TrtmntPsV TrtmntPr-C TrtmntPrV
TrtmntPst-C -0.520                                          
TrtmntPstVc -0.520  0.500                                   
TrtmntPr-Ch -0.520  0.500      0.500                        
TrtmntPrVcc -0.520  0.500      0.500     0.500              
StatusVacc  -0.632  0.000      0.000     0.000      0.000 

If I choose a different ASV to use as a response, the fixed effect values change considerably, but under correlation of fixed effects, only the intercept changes slightly, while the remaining values are all 0.500 or 0.000

[AlexisDBV]

Hello! I'm also running into this same issue where my LFCs are all 0, SEs are all 0.5, W are all 0, and p/q values are 1 only when I add a random formula to the model. I was wondering if there was a solution to this yet. My sessionInfo() is found below.

sessionInfo()
R version 4.2.3 (2023-03-15)
Platform: x86_64-apple-darwin17.0 (64-bit)
Running under: macOS 14.3.1

Matrix products: default
LAPACK: /Library/Frameworks/R.framework/Versions/4.2/Resources/lib/libRlapack.dylib

locale:
[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
[1] ANCOMBC_2.0.3   phyloseq_1.41.1

loaded via a namespace (and not attached):
  [1] readxl_1.4.3                   backports_1.4.1                Hmisc_5.1-1                    plyr_1.8.9                     igraph_1.5.0                  
  [6] lazyeval_0.2.2                 splines_4.2.3                  gmp_0.7-1                      BiocParallel_1.32.6            scater_1.26.1                 
 [11] GenomeInfoDb_1.34.9            ggplot2_3.5.0                  TH.data_1.1-2                  digest_0.6.34                  foreach_1.5.2                 
 [16] yulab.utils_0.1.4              htmltools_0.5.7                viridis_0.6.5                  lmerTest_3.1-3                 fansi_1.0.6                   
 [21] magrittr_2.0.3                 checkmate_2.3.1                memoise_2.0.1                  ScaledMatrix_1.6.0             cluster_2.1.6                 
 [26] doParallel_1.0.17              DECIPHER_2.26.0                Biostrings_2.66.0              matrixStats_1.2.0              sandwich_3.1-0                
 [31] colorspace_2.1-0               blob_1.2.4                     ggrepel_0.9.5                  rbibutils_2.2.16               xfun_0.42                     
 [36] dplyr_1.1.4                    crayon_1.5.2                   RCurl_1.98-1.14                jsonlite_1.8.8                 Exact_3.2                     
 [41] lme4_1.1-35.1                  survival_3.5-7                 zoo_1.8-12                     iterators_1.0.14               ape_5.7-1                     
 [46] glue_1.7.0                     gtable_0.3.4                   zlibbioc_1.44.0                emmeans_1.10.0                 XVector_0.38.0                
 [51] DelayedArray_0.24.0            BiocSingular_1.14.0            Rhdf5lib_1.20.0                SingleCellExperiment_1.20.1    Rmpfr_0.9-2                   
 [56] BiocGenerics_0.44.0            scales_1.3.0                   mvtnorm_1.2-4                  DBI_1.2.2                      rngtools_1.5.2                
 [61] Rcpp_1.0.12                    viridisLite_0.4.2              xtable_1.8-4                   htmlTable_2.4.2                decontam_1.18.0               
 [66] tidytree_0.4.6                 rsvd_1.0.5                     foreign_0.8-86                 bit_4.0.5                      proxy_0.4-27                  
 [71] Formula_1.2-5                  stats4_4.2.3                   CVXR_1.0-12                    htmlwidgets_1.6.4              httr_1.4.7                    
 [76] scuttle_1.8.4                  pkgconfig_2.0.3                nnet_7.3-19                    utf8_1.2.4                     tidyselect_1.2.0              
 [81] rlang_1.1.3                    reshape2_1.4.4                 munsell_0.5.0                  cellranger_1.1.0               tools_4.2.3                   
 [86] cachem_1.0.8                   cli_3.6.2                      DirichletMultinomial_1.40.0    RSQLite_2.3.5                  generics_0.1.3                
 [91] mia_1.6.0                      ade4_1.7-22                    evaluate_0.23                  biomformat_1.26.0              stringr_1.5.1                 
 [96] fastmap_1.1.1                  knitr_1.45                     bit64_4.0.5                    fs_1.6.3                       purrr_1.0.2                   
[101] rootSolve_1.8.2.4              sparseMatrixStats_1.10.0       nlme_3.1-164                   doRNG_1.8.6                    compiler_4.2.3                
[106] rstudioapi_0.15.0              beeswarm_0.4.0                 e1071_1.7-14                   treeio_1.22.0                  tibble_3.2.1                  
[111] DescTools_0.99.50              stringi_1.8.3                  gsl_2.1-8                      lattice_0.22-5                 Matrix_1.5-3                  
[116] nloptr_2.0.3                   vegan_2.6-4                    permute_0.9-7                  multtest_2.54.0                vctrs_0.6.5                   
[121] pillar_1.9.0                   lifecycle_1.0.4                rhdf5filters_1.10.1            BiocManager_1.30.22            Rdpack_2.6                    
[126] BiocNeighbors_1.16.0           estimability_1.4.1             irlba_2.3.5.1                  data.table_1.15.0              bitops_1.0-7                  
[131] lmom_3.0                       GenomicRanges_1.50.2           R6_2.5.1                       gridExtra_2.3                  vipor_0.4.7                   
[136] IRanges_2.32.0                 gld_2.6.6                      codetools_0.2-19               boot_1.3-30                    energy_1.7-11                 
[141] MASS_7.3-60.0.1                TreeSummarizedExperiment_2.6.0 rhdf5_2.42.1                   SummarizedExperiment_1.28.0    withr_3.0.0                   
[146] multcomp_1.4-25                S4Vectors_0.36.2               GenomeInfoDbData_1.2.9         MultiAssayExperiment_1.24.0    mgcv_1.9-1                    
[151] expm_0.999-7                   parallel_4.2.3                 beachmat_2.14.2                grid_4.2.3                     rpart_4.1.23                  
[156] tidyr_1.3.1                    coda_0.19-4.1                  class_7.3-22                   minqa_1.2.6                    DelayedMatrixStats_1.20.0     
[161] rmarkdown_2.26                 MatrixGenerics_1.10.0          numDeriv_2016.8-1.1            Biobase_2.58.0                 base64enc_0.1-3

[TonimaRahman]

@Freemacl
Hi, did you find any solution to the problem? I have a similar dataset and running into same problem

[Freemacl]

Unfortunately I did not

…

On Wed, Apr 30, 2025 at 10:53 PM TonimaRahman ***@***.***> wrote: @Freemacl <https://github.com/Freemacl> Hi, did you find any solution to the problem? I have a similar dataset and running into same problem — Reply to this email directly, view it on GitHub <#232 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACCWXMPUUJYZJOBFQTO2ZW324GSFJAVCNFSM6AAAAAB4HGZ6OSVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTGMBQGA4TCNY> . You are receiving this because you were mentioned.Message ID: ***@***.*** com>