Navigate to the config directory of MINUUR and open the file config.yaml. You will come across a list of options of how you want the pipeline to run. Let's go through these options.
samples: ../config/samples_list.tsv
input_type:
fastq: True
bam: False
input_config:
automatic: False
manual: ../config/sample_table.tsv
Samples
Do not edit this - this will point to your sample_list.tsv file in which you'll have listed your sample names.
input_type
Choose between fastq or bam. Most likely your input will be a fastq file. Some of the samples I processed were bam files so I needed this option to start the run at this point. Note that if you have a mixture of bam and fastq files, MINUUR will not be able to process both simultanesouly. Please do them one at a time.
input_config
Choose how you want to specify your files. automatic or manual. To understand the difference between the two, look back at the previous section.
kraken_db_location
kraken_db_location: /home/db/Kraken2_db/
Provide a path to where you've put your indexed Kraken2 database.
QC
QC:
Activate: True
CutadaptParams: "--minimum-length 50 -q 30"
CutadaptThreads: 10
Activate: Insert True if you would like to perform QC on your samples, False if you would like to skip.
CutadaptParams: Insert specific cutadapt parameters. For my reads I trimmed each read <50 bp and <30 phred score
CutadaptThreads: specify the number of threads you'd like to use
Remove Host From Fastq Gz
RemoveHostFromFastqGz:
Threads: 15
# Give path to directory - allows MINUUR to check there is an existing directory
ContaminantDirectory: /home/db/User_dbs/aidan_foo/bt_db/
# Give path to Bowtie2 indexed host genome
ContaminantIndex: /home/db/User_dbs/aidan_foo/bt_db/aegypti_db
AlignmentSensitivty: "--sensitive-local"
Threads: Specify the number of threads for bowtie2 to use to separate your indexed host reference
ContaminantDirectory: Specify the directory your indexed reference database is stored. This is used as a CheckPoint for the pipeline to confirm you have a host to separate
ContaminantIndex: Specify the path to your bowtie2 indexed host to separate. You will have a list of host.1.bt1, host.2.bt2, host.3.bt2, host.4.bt2, host.rev.1.bt2, host.rev.2.bt2. When specify this path, do so like /path/to/your/host - exclude the extension.
AlignmentSensitivity: specify the sensitivity and alignment type to align your reads. Options for this include
- "--very-fast-local"
- "--fast-local"
- "--sensitive-local"
- "--very-sensitive-local"
- "--very-fast-global"
- "--fast-global"
- "--sensitive-global"
- "--very-sensitive-global"
ProcessBam:
Activate: True
# Activate if input was Fastq (prior). If Input is bam then FromFastq = False
FromFastq: True
Activate: Pick True if you would like to generate some statistics of your alignments.
FromFastq: Pick True if you input fastq files. If you had bam files choose False
After QC, you will have some gzipped fastq files pertaining to reads that didn't map to your host in question. Cool! Lets see how much microbial information we can get from this. One path you can take is to classify those reads against the Kraken2 database you downloaded.
Kraken Classification
KrakenClassification:
Activate: True
Threads: 20
ConfidenceScore: 0
Activate : Pick True if you would like to do some taxonomic classifications with Kraken2
Threads: Choose the number of threads Kraken should use for classification
ConfidenceScore: Choose a number between 0 - 1 which sets how stringent the Kraken2 classifications will be. I picked 0, the default. 1 = very stringent.
Kraken Summaries
KrakenSummaries:
Activate: True
GenusReadThreshold: 6000
SpeciesReadThreshold: 10000
StratThreshold: 10000
KrakenDbStandard: True
Activate: Use True to get some summary data of your Kraken2 classifications. These can be more easily parsed and wrangled in R - especially in the Tidyverse.
GenusReadThreshold: MINUUR will attempt to make some plots of genus level classifications for you. Set a threshold for your plots so they don't appear to crowded.
SpeciesReadThreshold: Same as above but for species level classification.
StratThreshold: Again, a read threshold when making stratified plots of all your classifications.
Extract Kraken Classified Reads
This option is useful if you are interested in a specific taxon or just want to filter out bacterial / archael / viral reads. I've found this option useful when I wanted to do some metagenome assemblies but found a lot of reads unclassified. This could mean two things 1. There are reads left over from your host or 2. There are potentially mirobes present in your unmapped reads not present in the database. You can choose if you'd like to filter out those bacterial reads and use the resultant paired fastq files for metagenome assembly and binning in the later steps
ExtractKrakenTaxa:
Activate: True
taxon_choice: "2"
Activate: Use True to activate this option
taxon_choice: Use "2" to filter all bacteria. This is the taxonomy ID for bacteria
Bracken Reestimation
Bracken is used to redistribute Kraken2 classified reads to estimate taxonomic abundance in your sample. This will help remove some of the over/under estimation of classified reads at species and genus level.
BrackenReestimation:
Activate: True
BrakenDb: /home/db/Kraken2_db/
ClassificationLvl: 'S'
DistributionThresh: 10
PlotThreshold: 10000
Activate: Pick True if you would like to use Bracken to reestimate reads
BrackenDb: Give a path to your Bracken Database. FYI - if you downloaded the Kraken2 index from the link given prior, a bracken database is also included here, so specify the same path to your Kraken2 database.
ClassificationLvl: choose S for species or G for genus. Bracken will estimate taxonomic abundance at these levels accordingly.
PlotThreshold: MINUUR will try and make some plots for you - set a read threshold so the plots don't appear to crowded.
MetaPhlAn3 Classification
An alternative to Kraken2 is MetaPhlAn3. MetaPhlAn3 uses a marker gene appraoch for read classification. I found MetaPhlAn3 to perform poorly compared to Kraken2 for my mosquito metagenomic data.
MetaphlanClassification:
Activate: True
Database: /home/db/User_db/metaphlan_db
CleanMetaphlanReport: True
NProc: 8
Activate: Choose True to activate MetaPhlAn
Database: Give a path to your metaphlan database
CleanMetaphlanReport: Produce a summary file of your MetaPhlAn classifications to easily parse
NProc: Set the number of threads
The third path MINUUR takes is to try and do some metagenome assemblies, followed by binning of the resultant contigs and metagenome assembled genome QC. This was of most interest to me (and maybe you too!) as it meant I could recover MAGs associated to mosquitoes for further analysis. However...this won't always work. Given that the reads we're trying to assemble were never intended for metagenome assemblies it is possible you'll never get a set of high-quality MAGs. Nevertheless it's still worth trying! And I've had good success using this approach, especailly when the number of classified reads are high enough for a particular taxon. Around ~200,000 worked for me.
Metagenome Assembly
MetagenomeAssm:
Activate: True
UseKrakenExtracted: False
Threads: 14
Memory: 0.5
Activate: If True will perform metagenome assemblies with MegaHit on your paried unmapped fastq files.
UseKrakenExtracted: Another way to do your assemblies. If you extracted some taxa of interest, or extracted all bacterial classified reads, you can choose to do this option and perform metagenome assemblies on those reads. I found that assemblies with Kraken2 classified reads result in fewer MAGs but had less contamination and higher-completeness. Check out my paper for more info.
Binning
Binning is where we can pull some actual genomes from our assembeld contigs. MINUUR uses MetaBat2 to do this.
MetagenomeBinning:
Activate: True
Threads: 20
MinimumContigLength: 1500
Activate: Use True if you'd like to bin your genomes
Threads: set the number of threads to use
MinimumContigLength: Don't use contigs below a given threshold. The minimum here is 1500bp. If you go lower than this, you will raise an error.
Bin Quality Assurance
Check the quality of those bins! MINUUR will use CheckM to assess bin quality
CheckmBinQA:
Activate: True
Threads: 15
Activate: Use True to activate
Threads: specify the number of threads
Optionally you can also use BUSCO to assess bin quality
BUSCO:
Activate: True
Activate: Use True to activate
This can help identify whether low quality MAGs pertain to any eukaryotic type assemblies
By now you should a configured pipeline, with all the things you want to run and paths specified to all your databases. Lets run it!