DIfferentially abundant taxa but some of them are not present in both groups (DESeq2)
1
0
Entering edit mode
Valentín • 0
@2c7d2283
Last seen 9 weeks ago
Chile

Hello, I have an experiment and a control group to test differentially abundant taxa from amplicon sequencing data.

The problem is that after filtering by adjusted p-value < 0.01 , some of the differentially abundant OTUs are not present in both datasets, which is a problem for downstream analysis. I need to have the same taxa for both groups. ¿WHy do I have significant differentially abundant OTUs that are not present in both datasets? I thought the program internally only computes differences from taxa that are present in both groups.


design <- ~ age + asthma_rhinitis

dds_oral_data <- phyloseq_to_deseq2(merged_oral, design = design)

wald_test_merged_oral <- DESeq(dds_oral_data, parallel = TRUE)

results_oral_NO_AR <- results(wald_test_merged_oral, contrast = c("asthma_rhinitis", "AR","NO"))

# Extract the significant differences
df_diff_expressed_ARNO <- results_oral_NO_AR %>%
as.data.frame() %>%
rownames_to_column(var = "ASV") %>%
as_tibble() %>%

plseq_oral_AR <- prune_taxa(df_diff_expressed_ARNO$ASV, plseq_oral_AR) plseq_oral_NO <- prune_taxa(df_diff_expressed_ARNO$ASV, plseq_oral_NO)


the above two last variables are the respective phyloseq objects of the contrasted groups and both have different number of ASVs and they should have the same number of taxa given the fact that a differential abundance test was performed. How to deal with this?

DESeq2 • 139 views
0
Entering edit mode
@mikelove
Last seen 1 day ago
United States

You can just filter to ensure you only look at rows that have minimal counts in all groups that you require.

If you have a design of ~batch + condition, and there are no counts from one batch but clearly differential counts in the other group, you have evidence against the null that the condition has no effect on counts.