# melonnPan questions

**URL:** <https://forum.biobakery.org/t/melonnpan-questions/409>\
**Category:** MelonnPan\
**Created:** [April 29, 2020, 11:19pm UTC](https://forum.biobakery.org/t/melonnpan-questions/409 "2020-04-29T23:19:32Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![sagunmaharjann](https://yyz1.discourse-cdn.com/flex027/user_avatar/forum.biobakery.org/sagunmaharjann/32/168_2.png) [@sagunmaharjann](https://forum.biobakery.org/u/sagunmaharjann)\
**Post date:** [April 29, 2020, 11:19pm UTC](https://forum.biobakery.org/t/melonnpan-questions/409/1 "2020-04-29T23:19:32Z")

</div>

I recently read your paper “Predictive metabolomic profiling of microbial communities using amplicon or metagenomic sequences”, I was very impressed by the work done, and I am sure that melonnPan is a pivotal resource in the field of microbiome modeling. Being interested in it for my work, after reading the documentation and the tutorial, I decided to try obtaining some predictions with it. I was wondering if you could have time to help me with a couple of questions (or redirect me to whom to ask).

1. In melonnpan.train what is exactly the train.metag input? I could not fully understand from the function description
2. To obtain predictions with a previously trained model (different from the default one and obtained using melonnpan.train) is it enough to input a new weight matrix obtained using the melonnpan.train function?

---

<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:** [April 30, 2020, 12:23pm UTC](https://forum.biobakery.org/t/melonnpan-questions/409/2 "2020-04-30T12:23:56Z")

</div>

Hi @sagunmaharjann - train.metag is the input metagenome file (e.g. gene families as profiled by HUMAnN2) which MelonnPan uses as predictors. To answer your second question, yes, it is sufficient to supply a new weight to get predictions from a newly trained model. Thanks!

---

<div class="post-metadata">

**Author:** ![saif](https://yyz1.discourse-cdn.com/flex027/user_avatar/forum.biobakery.org/saif/32/377_2.png) [@saif](https://forum.biobakery.org/u/saif)\
**Post date:** [September 14, 2020, 5:10pm UTC](https://forum.biobakery.org/t/melonnpan-questions/409/3 "2020-09-14T17:10:22Z")

</div>

Hi @himel.mallick and @sagunmaharjann, I asked a similiar question in another post [MelonnPan training -- no common IDS](https://forum.biobakery.org/t/melonnpan-training-no-common-ids/792/5)  
Could you please see? I am getting an error message if I input a new weight obtained from melonnpan.train()

---

<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 16, 2020, 1:53pm UTC](https://forum.biobakery.org/t/melonnpan-questions/409/4 "2020-09-16T13:53:50Z")

</div>

@saif I just replied to the other thread. Let’s follow up either there or we can follow up here. Thanks!