---
layout: page
title: Sorting the unknown
---

We defined initially four categories of _unknowns_ (might be more in the future) trying to combine an ecological and a protein domain based approach to their definition. The categories are defined as follow:

-   **KNOWN**: Our knowns are all those ORFs that contains a Pfam domain. We are developing an approach to assign function to the unknown ORFs that relies on Domain Co-cocurrence Networks and uses Pfam as a basic building block.

-   **GENOMIC UNKNOWNS**: The first categories of unknowns are those ORFs with unknown function but associated to a sequenced organism, or to population genomes (aka Metagenome Assembled Genomes).

-   **ENVIRONMENTAL UNKNOWNS**: The second category of unkwnowns are those  ORFs with unknown function, which cannot be associated to an organism and are found only in environmental metagenomes

<div class="img_container" style="width:50%; margin:2em auto;">
<img alt="cl_categories.png" src="/img/cl_categories.png" width="80%">
</div>

<br />
<h2 class="section-heading  text-primary">A bioinformatic workflow to structure the unknown functional space</h2>

We have implemented a bioinformatic workflow that performs the partitioning of genomic metagenomic datasets on the different categories of KNOWNS and UNKNOWNS.

<div class="img_container" style="width:50%; margin:2em auto;">
<img alt="methodology.png" src="/img/methodology.png" width="80%">
</div>

We start from a de-novo clustering of all genomic and environmental genes and continue through a complex pipeline that validates and characterizes the gene clusters. For a more detailed explanation of the pipeline check how we [create](3_MMseqs_clustering) the protein clusters, how we do the [validation](5_Cluster_validation) and how we [classify](7_Cluster_classification) them.