how to export the file of the contribution of each ASV to the abundance of each predicted gene family.

[CamilleZhao7]

after runing picrust2_pipeline.py
-s dna-sequences.fasta
-i feature-table.biom
-o picrust2_output_pipeline
-p 36
I got the pathway_out > path_abun_unstrat.tsv.gz how to get the contribution of each ASV to the abundance of each predicted gene family.

[cafferychen777]

Hi @CamilleZhao7,

Great question! The path_abun_unstrat.tsv.gz file gives you the total pathway abundances across all ASVs, but to see the individual ASV contributions to gene families, you need to look at the intermediate files from the PICRUSt2 pipeline.

Finding ASV Contributions to Gene Families

The information you're looking for is in the stratified gene family output, which should be located in your PICRUSt2 output directory. Here's how to access it:

1. Check your PICRUSt2 output structure:

ls -la picrust2_output_pipeline/

You should see directories like:

• EC_metagenome_out/ (EC predictions)
• KO_metagenome_out/ (KEGG Ortholog predictions)
• pathways_out/ (pathway predictions)

2. Look for stratified gene family files:

The key files you want are:

• KO_metagenome_out/pred_metagenome_strat.tsv.gz - Shows KO gene family abundances stratified by each ASV
• EC_metagenome_out/pred_metagenome_strat.tsv.gz - Shows EC gene family abundances stratified by each ASV

3. Extract and examine the stratified data:

# Decompress the stratified KO file
gunzip -c picrust2_output_pipeline/KO_metagenome_out/pred_metagenome_strat.tsv.gz > ko_strat.tsv

# Look at the first few lines
head ko_strat.tsv

4. Understanding the stratified format:

The stratified file has this structure:

function	ASV_ID|Taxon	Sample1	Sample2	Sample3	...
K00001	ASV_1|s__Species_name	10.5	15.2	8.1	...
K00001	ASV_2|s__Species_name	5.3	12.8	3.7	...
K00002	ASV_1|s__Species_name	8.9	11.4	6.2	...

Each row shows the abundance of a specific gene family (e.g., K00001) contributed by a specific ASV across all your samples.

5. Processing in R (using ggpicrust2):

If you want to analyze this data with ggpicrust2, you can load the stratified data:

library(ggpicrust2)

# Read the stratified KO data
ko_strat <- read.table("ko_strat.tsv", sep="\t", header=TRUE, row.names=1, check.names=FALSE)

# The row names will be in format: "function|ASV_ID|Taxon"
# You can separate the function and ASV information
rownames_split <- strsplit(rownames(ko_strat), "|", fixed=TRUE)
functions <- sapply(rownames_split, function(x) x[1])
asvs <- sapply(rownames_split, function(x) x[2])

# Create a data frame with function, ASV, and abundance data
strat_data <- data.frame(
  function_id = functions,
  asv_id = asvs,
  ko_strat,
  stringsAsFactors = FALSE
)

6. Alternative: Using STAMP or other tools

You can also import the stratified files into tools like:

• STAMP (Statistical Analysis of Metagenomic Profiles)
• LEfSe (Linear discriminant analysis Effect Size)
• Custom R/Python scripts for further analysis

Summary

• ✅ path_abun_unstrat.tsv.gz = Total pathway abundances (what you have)
• ✅ pred_metagenome_strat.tsv.gz = Individual ASV contributions to gene families (what you need)
• ✅ The stratified files show exactly which ASVs contribute how much to each predicted gene family

The stratified output is the key to understanding the microbial community structure behind your functional predictions - it tells you not just what functions are present, but which microbes are contributing to those functions.

Hope this helps! Let me know if you need help processing or analyzing the stratified data further.

Best regards,
ggpicrust2 team