Error in ggpicrust2 (There are no statistically significant biomarkers in the dataset)

[erda712]

Dear Chen, Thank you very much for creating this useful analysis tool. I am starting to use the package to analyse my PICRUST2 result, however I have got this error in my analysis: Error in ggpicrust2(data = abundance_data, metadata = metadata_Nepal_kath, : Notice: There are no statistically significant biomarkers in the dataset. This is not an error, but it might indicate that the data do not contain any biomarkers passing the set significance threshold (p<=0.05). You may refer to the tutorial's FAQ for further help and suggestions. For your information below is the codes I used following tutorial in: https://cran.rproject.org/web/packages/ggpicrust2/vignettes/using_ggpicrust2.html:

abundance_file <- "KO.tsv" metadata_Nepal_kath <- read.csv("metadata_perimilk_kath.csv") results_data_input <- ggpicrust2(data = abundance_data, metadata = metadata_Nepal_kath, group = "ABRpresence", pathway = "KO", daa_method = "LinDA", ko_to_kegg = TRUE, order = "pathway_class", p_values_bar = TRUE, x_lab = "pathway_name")

LinDA analysis is complete.

Error in ggpicrust2(data = abundance_data, metadata = metadata_Nepal, :

Notice: There are no statistically significant biomarkers in the dataset. This is not an error, but it might indicate that the data do not contain any biomarkers passing the set significance threshold (p<=0.05). You may refer to the tutorial's FAQ for further help and suggestions.

In addition: Warning message:

In MicrobiomeStat::linda(abundance, LinDA_metadata_df, formula = "~Group_groupnonsense", :

Some features have less than 3 nonzero values!

They have virtually no statistical power. You may consider filtering them in the analysis!

I have tried to do it with different group from the metadata, but still it gave the same error.

I really hope that you can help me with this issue, what could be the possible reasons and what should I do. Thank you very much Chen, your help is much appeciated. Best Regards Erda

[cafferychen777]

Dear @erda712,

Thank you for reaching out and using ggpicrust2 for your analysis. I understand your concern regarding the absence of statistically significant biomarkers in your dataset. This situation is not uncommon, and there are several reasons why it might occur:

1. True Difference Between Groups: If the microbial communities or pathways you're comparing are very similar, it's possible that there genuinely are no significant differences between them.

2. Sample Size: A small sample size can limit your statistical power, making it difficult to detect differences even if they exist.

3. Intra-Group Variation: High variation within your groups can also obscure differences between them. This could be due to outliers or subgroups that increase variability.

To address these issues, you might consider the following:

• Increase Sample Size: If possible, adding more samples can improve your statistical power.

• Decrease Intra-Group Variation: Check for outliers or subgroups that might be driving the variation. Cleaning your data or stratifying your analysis could help.

• Adjust Statistical Methods or Parameters: Different statistical methods or parameter adjustments might be more suitable for your data. For example, switching between parametric and non-parametric tests or tweaking the parameters of your current test.

It's important to remember that not finding significant results is still a valid outcome. It might indicate that there are no substantial differences between the groups you're studying. Always interpret your results in the context of your specific study.

For more detailed suggestions and troubleshooting, please refer to the FAQ section in our GitHub repository.

I hope this helps! If you have any further questions or need more assistance, please don't hesitate to reach out.

Best regards, Chen

[erda712]

Thank you Chen for your prompt reply. OK, I'll try work on it, and will give updates about it. Thank you.

Best Regards Erda

[erda712]

Dear Chen, I would like ask for more assistance please. I have tried different ways as suggested in GitHub but I still got the same error. If I could ask for your help to look at my data and see if it gives the same error if you perform the analysis then it would be really helpful for me. Thank you very much Chen, appreciate your assistance. Best Regards Erda

[cafferychen777]

Dear @erda712,

I'm sorry to hear you are still getting the same error after trying different approaches as I suggested. This error typically means that there are no statistically significant differences in microbial abundances between your groups in the analysis. A few things to check:

1) Make sure your groups have sufficient sample size (at least 3-5 samples per group). Too few samples can limit statistical power to detect differences.

2) Check that your groups are truly biologically distinct - for example, different disease states, treatment conditions, etc. Groups without real differences will not show any significant differences in analysis.

3) Double check your metadata - make sure samples are assigned to the proper groups, no typos or errors.

If you still see no statistical differences after careful checking, then it likely reflects a lack of strong microbiome differences between your particular groups. I'm happy to take a closer look if you want to share some example data files with me. Just send over some example abundance and metadata files and I can run the analysis on my end to see if I find the same issue.

Let me know if you would like me to take a look or if you have any other questions!

Best regards, Chen YANG