# Metabolite Prediction using 16S microbiome data

**URL:** <https://forum.biobakery.org/t/metabolite-prediction-using-16s-microbiome-data/261>\
**Category:** MelonnPan\
**Created:** [March 7, 2020, 8:37pm UTC](https://forum.biobakery.org/t/metabolite-prediction-using-16s-microbiome-data/261 "2020-03-07T20:37:30Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![Aditya\_Bajaj](https://yyz1.discourse-cdn.com/flex027/user_avatar/forum.biobakery.org/aditya_bajaj/32/142_2.png) [@Aditya\_Bajaj](https://forum.biobakery.org/u/Aditya_Bajaj)\
**Post date:** [March 7, 2020, 8:37pm UTC](https://forum.biobakery.org/t/metabolite-prediction-using-16s-microbiome-data/261/1 "2020-03-07T20:37:30Z")

</div>

Hello. We tried to perform the melonnpan-based metabolite prediction on our 16s sequencing raw data and we have come across a few doubts.

1. Does Melonnpan accept the output from Picrust or from Humann2?
2. If Melonnpan accepts Picrust output only, then how do we convert the KO gene IDs (produced by Picrust) to their respective Uniref90 IDs (input format of Melonnpan)?
3. We tried giving our Picrust output to Humann2 to generate Uniref IDs, but Humann2 is only able to produce pathway abundances via a keggc- adducted route.  
How can we bridge this gap in the Picrust (or Humann2) output files and the melonnpan input specifications?  
Thanks  
Aditya

---

<div class="post-metadata">

**Author:** ![himel.mallick](https://yyz1.discourse-cdn.com/flex027/user_avatar/forum.biobakery.org/himel.mallick/32/1757_2.png) [@himel.mallick](https://forum.biobakery.org/u/himel.mallick)\
**Post date:** [September 4, 2020, 1:57pm UTC](https://forum.biobakery.org/t/metabolite-prediction-using-16s-microbiome-data/261/2 "2020-09-04T13:57:11Z")

</div>

Hi @Aditya_Bajaj - this topic has been discussed here: [MelonnPan Input - Picrust2 output](https://forum.biobakery.org/t/melonnpan-input-picrust2-output/267/2).

In short, currently, it is not possible to use Picrust/Picrust2 KO output with the default MelonnPan-Predict module. However, the users can still use the MelonnPan-Train module to train a new model with KOs derived from Picrust/Picrust2.

---

<div class="post-metadata">

**Author:** ![Zubair2021](https://yyz1.discourse-cdn.com/flex027/user_avatar/forum.biobakery.org/zubair2021/32/1849_2.png) [@Zubair2021](https://forum.biobakery.org/u/Zubair2021)\
**Post date:** [December 27, 2022, 10:06am UTC](https://forum.biobakery.org/t/metabolite-prediction-using-16s-microbiome-data/261/3 "2022-12-27T10:06:28Z")

</div>

Thank you @himel.mallick for your suggestions. Since you mentioned training a new model, do I need to have metagenomic data paired with metabolomic data? What could be a potential source of these inputs for a chicken gut microbiome?

---

<div class="post-metadata">

**Author:** ![himel.mallick](https://yyz1.discourse-cdn.com/flex027/user_avatar/forum.biobakery.org/himel.mallick/32/1757_2.png) [@himel.mallick](https://forum.biobakery.org/u/himel.mallick)\
**Post date:** [December 27, 2022, 6:42pm UTC](https://forum.biobakery.org/t/metabolite-prediction-using-16s-microbiome-data/261/4 "2022-12-27T18:42:32Z")

</div>

Hi @Zubair2021 - unfortunately, I am not sure about a potential source of these inputs for a chicken gut microbiome and I am afraid without paired metagenomic and metabolomic data, you won’t be able to use MelonnPan as currently implemented.

Best regards,  
Himel