Number of samples needed for accuracy

Microbial community profiling FUGAsseM
1

Hello,

I was wondering if you have any sense of how many metatranscriptomic/metagenomic samples would be needed to train FUGASSEM for it be accurate enough to make usable predictions.

Thanks

2

Hi there,

Thanks for your question! The short answer is that it depends on diversity and the functions you want to predict, but there are some practical guidelines for your information:

  1. 100~300 metagenomes (MGX) paired with metatranscriptomes (MTX) can yield usable signals for common functions, provided samples span the major community states you care about.
  2. 500~1,000 MGX and MTX pairs typically stabilize performance and reduce variance across sites/batches.

Best,
Yancong