https://github.com/silask/krak

Snakemake for flexitaxd and Kraken2
https://github.com/silask/krak

Last synced: 2 months ago
JSON representation

Snakemake for flexitaxd and Kraken2

• Host: GitHub
• URL: https://github.com/silask/krak
• Owner: SilasK
• Created: 2021-01-11T14:21:37.000Z (over 4 years ago)
• Default Branch: master
• Last Pushed: 2022-08-12T15:16:02.000Z (almost 3 years ago)
• Last Synced: 2025-04-10T06:07:54.829Z (2 months ago)
• Language: Python
• Size: 63.5 KB
• Stars: 6
• Watchers: 3
• Forks: 3
• Open Issues: 3
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Simple snakemake pipline for Kraken2 + Braken.

Running kraken+ braken can be a bit combersome. Here is *Krak* a simple plpeline that let's you run these tools in a simple and efficient way. 

## Classify using existing kraken2 + braken dbs

Download git repository:

    git clone https://github.com/SilasK/Krak

Install Snakemake using [conda](https://conda.io/projects/conda/en/latest/user-guide/install/index.html):

    conda create -c bioconda -c conda-forge -n kraken snakemake

    conda activate kraken

For installation details, see the [instructions in the Snakemake documentation](https://snakemake.readthedocs.io/en/stable/getting_started/installation.html).

2. generate a sample table from your fastq files:

        krak/scripts/generate_sample_table.py path/to/fastqs

Check the generated samples.tsv and simplify names, if you wish.

3. Run the Pipeline.

a. First make a dryrun

      snakemake  -s krak/workflow/classify.smk --use-conda --dryrun

  If you have access to cluster or cloud we recommend you to setup cluster/cloud integration via [our profile](https://github.com/Snakemake-Profiles/generic). 

  See the [Snakemake documentation](https://snakemake.readthedocs.io/en/stable/executable.html) for further details.

  b. Run the Pipeline:

    snakemake  -s krak/workflow/classify.smk --use-conda 

## Use databases for human and mouse gut

Download the Kraken db for human and put them in `databases`:

    db="uhgg" # for human "cmmg" for mouse"

    mkdir -p databases/$db

    

    for file in  database100mers.kmer_distrib database150mers.kmer_distrib database200mers.kmer_distrib database250mers.kmer_distrib database50mers.kmer_distrib database75mers.kmer_distrib hash.k2d inspect.txt.gz opts.k2d seqid2taxid.map.gz taxo.k2d ; 

    do 

        wget https://ezmeta.unige.ch/CMMG/Kraken2db/$db/$file -O databases/$db/$file

    done

    

Set the path to the downloaded kraken db in the `Kraken/config.yaml`.

## Analyze the microbiome based on taxonomy and Kegg modules

Once you have the Kraken2 quantification you can use this [jupyter notebook](https://colab.research.google.com/github/trajkovski-lab/CMMG/blob/main/notebooks/Analyze-cold-adapted-microbiota.ipynb) that shows how CMMG can be used for the functional and taxonomic analysis of mouse metagenome data.

## Create custom kraken + braken dbs

Custom kraken db's can be generated using flexitaxd.

You need flextaxd to build Kraken dbs. There is an error in the latest version https://github.com/FOI-Bioinformatics/flextaxd/issues/48, therfore run

    mamba install flextaxd=0.3.5

    

    

 What you need to build a kraken db is:

  - path to the top-level-folder with all genomes

  - a greeen genes formated taxonomy file

Genomes not in the taxonaomy are ignored. 

  

Modify the name of your database as well as the path for building in the `config.file` 

The taxonomy file should be a tsv file of the fomrat:

```

GUT_GENOME000047        d__Bacteria;p__Proteobacteria;c__Gammaproteobacteria;o__Enterobacterales;f__Enterobacteriaceae;g__Enterobacter;s__Enterobacter mori

GUT_GENOME000565        d__Bacteria;p__Proteobacteria;c__Gammaproteobacteria;o__Enterobacterales;f__Enterobacteriaceae;g__Kluyvera;s__Kluyvera sp902363335

```

You can got down to subspecies level using the `x__` prefix

If you want to quantify on subspecies/strain level you should add `subsp` to the list of levels to be quantified.

## Genome length normalisation

This needs to be done. 

Kraken + Braken attributes reads to species (or the level you want) however it doesn't normalize it to genome size. Larger genomes get more reads. While this is no problem for ratio based analyses (CoDa) it biases relative abundance calculation.