Having issues with generate_taxa_test_long

[adahal123]

####FEATURE LEVEL ANALYSIS#######

This code works great
test.list <- generate_taxa_trend_test_long(
data.obj = ps.obj,
subject.var = "pid",
time.var = "visit",
group.var = "race",
adj.vars = "sex",
prev.filter = 0.0,
abund.filter = 0.0,
feature.level = c("Genus"),
feature.dat.type = c("proportion")
)

generate_taxa_trend_volcano_long(
data.obj = ps.obj,
group.var = "race",
time.var = "visit",
test.list = test.list,
feature.sig.level = 0.1,
feature.mt.method = "none"
)

This code below does not work:
result2 <- generate_taxa_test_long(
data.obj = ps.obj,
subject.var = "pid",
time.var = "visit",
group.var = "race",
adj.vars = "sex",
prev.filter = 0.0,
abund.filter = 0.0,
feature.level = c("Genus"),
feature.dat.type = c("proportion")
)

Results:

It then excludes samples from visit 2 and visit 3 and then shows this message:

As you can see, generate taxa trend works and so do the alpha diversity functions with these same variables specified. Sorry for asking so many questions.

[cafferychen777]

Hi @adahal123,

Thank you for bringing this issue to our attention. It appears that there might be a problem with the experimental design that is causing the generate_taxa_test_long function to exclude samples from visit 2 and visit 3.

To further investigate this, could you please try running your data object through the mStat_summarize_data_obj function with the appropriate arguments for time.var and group.var? For example:

summary_result <- mStat_summarize_data_obj(data.obj, time.var = "visit", group.var = "race")

This will help us understand if the issue is indeed related to the experimental design or if there's something else we need to address.

Please let me know how it goes, and if you have any further questions, feel free to ask.

Best regards,
Caffery Yang

[adahal123]

Ok this is for the alpha question which was in the other thread.
Using numeric for t0 and ts does not work. As you can see from my previous examples in the other thread, putting numeric for t0 and character values for ts worked for other functions but just not for generate_alpha_test_long

Still getting the same error message as before.

[cafferychen777]

Hi, thanks for reaching out again! It seems like the issue you're encountering with the generate_taxa_test_long function might be related to how the time variable is being handled in different functions. In other functions, if the time variable is character or factor, it could cause problems, so I've added a step to convert it to numeric (data.obj$meta.dat <- data.obj$meta.dat %>% dplyr::mutate(!!sym(time.var) := as.numeric(!!sym(time.var)))). However, for generate_alpha_test_long, the test is conducted separately for each time point, so the format of the time variable shouldn't be a problem, and I didn't add this conversion step.

This might explain why it works for other functions but not for generate_alpha_test_long. I believe specifying t0.level = 1 and ts.levels = c(2,3) should not cause issues in other functions, and I suggest doing the same in generate_alpha_test_long for consistency. Please let me know if this resolves the issue or if you have any further questions!

[adahal123]

Like I mentioned in my last post, specifying t0.level = 1 and ts.levels = c(2,3) does not work for generate_alpha_test_long does not work.

[cafferychen777]

Hi adahal123,

Thanks for reporting the issues you encountered with the generate_taxa_test_long and generate_alpha_test_long functions in the MicrobiomeStat package.

Regarding the problem with generate_taxa_test_long, I understand that it's not working for you even though other related functions like generate_taxa_trend_test_long and generate_taxa_trend_volcano_long are working fine with the same dataset and parameters. However, I've been unable to reproduce this error using the example data. Without access to the specific dataset you are working with, it's difficult for me to investigate why this issue is occurring in your case.

Similarly, for the generate_alpha_test_long function, specifying numeric values for the t0.level and ts.levels arguments doesn't seem to be working based on the error message you shared. But again, I couldn't replicate this behavior with the example data I have.

If you are able to share a small subset of your data that demonstrates these problems (either publicly or privately), that would be extremely helpful. Having the actual data will allow me to dig deeper, reproduce the errors, and hopefully figure out what's causing these functions to fail in your specific use case.

Please let me know if you can provide some sample data, and I'll be happy to take a closer look. Apologies for the inconvenience, and thanks in advance for your help in debugging these issues.

Best regards,
Caffery