A strange error when using MBA

renyiyuan · January 29, 2024, 12:50pm

AFNI version info (afni -ver): 23.3.14

Hi AFNI members:

When I use MBA to continue analyzing the results from 3dNetcorr, I encounter this strange error:

Loading required package: Rcpp
Loading 'brms' package (version 2.20.4). Useful instructions
can be found by typing help('brms'). A more detailed introduction
to the package is available through vignette('brms_overview').

Attaching package: ‘brms’

The following object is masked from ‘package:stats’:

    ar

Error in `$<-.data.frame`(`*tmp*`, ROI1, value = integer(0)) : 
  replacement has 0 rows, data has 18648
Calls: $<- -> $<-.data.frame
Execution halted

This is my script:

MBA -prefix ${group_dir}/exp_test -r2z -dataTable ${group_dir}/exp.table \
    -chains 4 -iterations 1000 -WCP 3 \
    -model '1+group' -EOI 'Intercept, group'

and the end of the input table:

or a sample table:

Subj,group,ROI1,ROI2,Y
sub-05_ses-before,before,EXA,GPe,0.24
sub-05_ses-before,before,EXA,GPi,0.1331
sub-05_ses-before,before,EXA,HN,-0.053
sub-05_ses-before,before,EXA,HTH,0.1292
sub-05_ses-before,before,EXA,LH_DefaultA_PFCd_1,0.1732
sub-05_ses-before,before,EXA,LH_DefaultA_PFCm_1,0.1122
sub-05_ses-before,before,EXA,LH_DefaultA_pCunPCC_1,0.0362
sub-05_ses-before,before,EXA,LH_DefaultB_IPL_1,0.2402
sub-05_ses-before,before,EXA,LH_DefaultB_PFCd_1,0.2737
sub-05_ses-before,before,EXA,LH_DefaultB_PFCl_1,0.1783
sub-05_ses-before,before,EXA,LH_DefaultB_PFCv_1,0.2059
sub-05_ses-before,before,EXA,LH_DefaultB_PFCv_2,0.2219
sub-05_ses-before,before,EXA,LH_DefaultB_Temp_1,0.1228
sub-05_ses-before,before,EXA,LH_DefaultB_Temp_2,0.1899
sub-05_ses-before,before,EXA,LH_DefaultC_PHC_1,0.1311
sub-05_ses-before,before,EXA,LH_DefaultC_Rsp_1,-0.0423
sub-05_ses-before,before,EXA,MN,0.0359
sub-05_ses-before,before,EXA,NAC,0.1941
sub-05_ses-before,before,EXA,PBP,-0.072
sub-05_ses-before,before,EXA,Pu,0.1985
sub-05_ses-before,before,EXA,RH_DefaultA_IPL_1,0.1556
sub-05_ses-before,before,EXA,RH_DefaultA_PFCd_1,0.2147
sub-05_ses-before,before,EXA,RH_DefaultA_PFCm_1,0.1349
sub-05_ses-before,before,EXA,RH_DefaultA_pCunPCC_1,0.0287
sub-05_ses-before,before,EXA,RH_DefaultB_PFCd_1,0.2306
sub-05_ses-before,before,EXA,RH_DefaultB_PFCv_1,0.0936
sub-05_ses-before,before,EXA,RH_DefaultB_PFCv_2,0.1512
sub-05_ses-before,before,EXA,RH_DefaultC_PHC_1,-0.0335
sub-05_ses-before,before,EXA,RH_DefaultC_Rsp_1,-0.0081
sub-05_ses-before,before,EXA,RN,0.0137
sub-05_ses-before,before,EXA,SNc,-0.0174
sub-05_ses-before,before,EXA,SNr,0.0175
sub-05_ses-before,before,EXA,STN,0.0017
sub-05_ses-before,before,EXA,VTA,-0.0475
sub-05_ses-before,before,EXA,VeP,0.3442
sub-05_ses-after,after,EXA,GPe,0.1789
sub-05_ses-after,after,EXA,GPi,0.0458
sub-05_ses-after,after,EXA,HN,-0.0743
sub-05_ses-after,after,EXA,HTH,0.1431
sub-05_ses-after,after,EXA,LH_DefaultA_PFCd_1,0.0322
sub-05_ses-after,after,EXA,LH_DefaultA_PFCm_1,0.1714
sub-05_ses-after,after,EXA,LH_DefaultA_pCunPCC_1,0.0687
sub-05_ses-after,after,EXA,LH_DefaultB_IPL_1,-0.0076
sub-05_ses-after,after,EXA,LH_DefaultB_PFCd_1,0.0061
sub-05_ses-after,after,EXA,LH_DefaultB_PFCl_1,-0.0363
sub-05_ses-after,after,EXA,LH_DefaultB_PFCv_1,0.033
sub-05_ses-after,after,EXA,LH_DefaultB_PFCv_2,0.0836
sub-05_ses-after,after,EXA,LH_DefaultB_Temp_1,0.0222
sub-05_ses-after,after,EXA,LH_DefaultB_Temp_2,0.0076
sub-05_ses-after,after,EXA,LH_DefaultC_PHC_1,0.0079
sub-05_ses-after,after,EXA,LH_DefaultC_Rsp_1,0.0724
sub-05_ses-after,after,EXA,MN,0.0929
sub-05_ses-after,after,EXA,NAC,0.0059
sub-05_ses-after,after,EXA,PBP,-0.1226
sub-05_ses-after,after,EXA,Pu,0.0105
sub-05_ses-after,after,EXA,RH_DefaultA_IPL_1,0.0829
sub-05_ses-after,after,EXA,RH_DefaultA_PFCd_1,0.0884
sub-05_ses-after,after,EXA,RH_DefaultA_PFCm_1,0.2084
sub-05_ses-after,after,EXA,RH_DefaultA_pCunPCC_1,0.0377
sub-05_ses-after,after,EXA,RH_DefaultB_PFCd_1,0.0746
sub-05_ses-after,after,EXA,RH_DefaultB_PFCv_1,0.1315
sub-05_ses-after,after,EXA,RH_DefaultB_PFCv_2,0.1071
sub-05_ses-after,after,EXA,RH_DefaultC_PHC_1,0.0782
sub-05_ses-after,after,EXA,RH_DefaultC_Rsp_1,0.0874
sub-05_ses-after,after,EXA,RN,-0.0787
sub-05_ses-after,after,EXA,SNc,0.0124
sub-05_ses-after,after,EXA,SNr,-0.0183
sub-05_ses-after,after,EXA,STN,-0.0022
sub-05_ses-after,after,EXA,VTA,0.02
sub-05_ses-after,after,EXA,VeP,0.1731

When I prepared the data for MBA, I didn't find some useful tutorial. So I did it in a hard way. First, I used fat_mvm_prep.py to generate the table. But that table could not be used directly by MBA. I needed to format it using Excel according to the MBA help. The whole process was so torturous. I hope some expert can give us more detailed tutorials about the preparatory stage of group analysis，like after running 3dNetcorr, you can use A to generate the input for B...

Gang · January 29, 2024, 3:13pm

The current version of MBA requires the input in long format as a data frame and does not support comma-separated values (CSV) directly. To adapt your file exp.table, consider transforming it into a data frame in R using the following command:

d <- read.csv('exp.table')
write.table(d, file = "new_exp.table", append = FALSE, quote = FALSE, row.names = FALSE)

Afterwards, you can use the newly created new_exp.table as input for MBA.

Gang Chen

renyiyuan · January 30, 2024, 7:48am

@Gang Thanks.
But another error occurred:

Syntax error in '/tmp/Rtmp297KX3/model_af58a03ed990050096d8fb59fd5a4e94.stan', line 22, column 5 to column 6, parsing error:
   -------------------------------------------------
    20:     *   an integer sequence from start to end
    21:     */
    22:    int[] sequence(int start, int end) {
              ^
    23:      array[end - start + 1] int seq;
    24:      for (n in 1:num_elements(seq)) {
   -------------------------------------------------

An identifier is expected as a function name.

Error: Syntax error found! See the message above for more information.
Execution halted

renyiyuan · January 30, 2024, 6:47pm

Hi @Gang,
After running 3dNetCorr, can I use part of that results as the input to RBA? Or use the ROI average of seed based whole brain correlation results as the input?

renyiyuan · January 31, 2024, 4:22am

@Gang This discussion may be useful for debugging WCP error in MBA and RBA.
https://discourse.mc-stan.org/t/new-array-syntax-inside-functions/32892

Gang · January 31, 2024, 10:41pm

@renyiyuan, it seems there is an issue with the R package brms. Here are two temporary solutions (either would work):

Avoid using the option -WCP in RBA and MBA.
Install older versions of cmdstanr and cmdstan using the following steps:

a. Install an old version of cmdstanr in R:
```
remotes::install_github("stan-dev/cmdstanr", ref = "v0.5.3")
```
b. Remove the current version of cmdstan from a location like $home/.cmdstan/cmdstan-x.x.x.

c. Install an old version of cmdstan in R:
```
library(cmdstanr)
install_cmdstan(cores = getOption("mc.cores", 2), version = '2.30.1')
```

I hope this helps resolve the issue.

Gang Chen

renyiyuan · February 1, 2024, 8:55am

@Gang , thanks! I have tried cmdstan-2.31.0. It works.

renyiyuan · February 1, 2024, 11:55am

@Gang ,
I check the RBA running log and see this error message:

Chain 1 Informational Message: The current Metropolis proposal is about to be rejected because of the following issue:
Chain 1 Exception: Exception: student_t_lpdf: Scale parameter is inf, but must be positive finite! (in '/tmp/RtmpJMya6K/model-1ee872a588f2e.stan', line 84, column 4 to column 62) (in '/tmp/RtmpJMya6K/model-1ee872a588f2e.stan', line 162, column 4 to line 164, column 51)
Chain 1 If this warning occurs sporadically, such as for highly constrained variable types like covariance matrices, then the sampler is fine,
Chain 1 but if this warning occurs often then your model may be either severely ill-conditioned or misspecified.

I don't know what it means. Can you explain it?

Gang · February 1, 2024, 9:18pm

I'm not entirely certain about the precise meaning of this warning, but I suspect it may not be a critical issue, as discussed here.

Gang Chen

renyiyuan · February 2, 2024, 10:42am

I advise adding a options in the R interface to block this low-level warning. And thanks a lot.

Gang · February 4, 2024, 1:44pm

To suppress the warning, install an older version of cmdstanr in R using the following command as suggested earlier:

remotes::install_github("stan-dev/cmdstanr", ref = "v0.5.3")

Gang Chen

renyiyuan · February 4, 2024, 2:36pm

Ok. Thanks. Compared to other issues, they are too trivial.

Gang · February 5, 2024, 12:54pm

The error and warning appear to have been resolved in the latest development version of brms, which you can install in R using the following commands:

devtools::install_github("stan-dev/cmdstanr")
install_cmdstan(cores = getOption("mc.cores", 4))
if (!requireNamespace("remotes")) {
  install.packages("remotes")
}
remotes::install_github("paul-buerkner/brms")

Gang Chen

renyiyuan · February 5, 2024, 2:08pm

OK. I will try it later.