miRNA

Necessary inputs

You will need to create a samplesheet with information about the samples you would like to analyse before running the pipeline. It has to be a comma-separated file with 5 columns, and a header row as shown in the examples below.

sample,fastq_1,fastq_2,condition,batch
CONTROL_REP1,AEG588A1_S1_L002_R1_001.fastq.gz,AEG588A1_S1_L002_R2_001.fastq.gz,control,1,0
CONTROL_REP1,AEG588A1_S1_L003_R1_001.fastq.gz,AEG588A1_S1_L003_R2_001.fastq.gz,control,2,0
CONTROL_REP1,AEG588A1_S1_L004_R1_001.fastq.gz,AEG588A1_S1_L004_R2_001.fastq.gz,control,3,0

Important

Sample names have to be in the first column or in a column called sampleID.

If you have column names other than condition and batch you need to declare the names in the params_mirna.yml. See below (Preprocess,DEA,PEA)

sampleID,condition
CONTROL_REP1,ctrl
CONTROL_REP2,ctrl
TREATMENT_REP1,treat

Start from SRA

Alternatively, instead of providing with paths of fastq files you can provide a column with SRA run identifiers you wish to download from NCBI in the first column named sample and the second column has to be named sampleID and store the SRA codes:

sample,sampleID,condition,batch,cl
SRR2015757,SRX1022901,control,1,1
SRR2015760,SRX1022904,control,2,1
SRR2015761,SRX1022905,control,3,1

An example samplesheet has been provided with the pipeline.

Reference files

The user can choose to run the pipeline using fasta and gtf files supplied by aws:

params{
  igenomes_base              : 's3://ngi-igenomes/igenomes'
  igenomes_ignore            : false
}

Alternatively, they can provide the location of local reference files:

params{
  fasta_mirna                   : '/home/bianca/gencode2/GRCh38.primary_assembly.genome.fa.gz'
  transcript_fasta_mirna         : '/home/bianca/gencode2/gencode.v44.transcripts.fa.gz'
  gtf_mirna                       : '/home/bianca/gencode2/gencode.v44.annotation.gtf.gz'

Running the pipeline

In order to run the miRNA part of the pipeline you have to modify one file, specifying which part of the analysis you want to run and specific parameters params_mirna.yml:

params{
genome = 'GRCh38'  # Reference genome identifier from AWS, check /conf/igenomes.config
outdir  = 'full path of location you want to output'
salmonDirmiRNA = 'full path of location you want to output/mirna/'
input_mirna = 'full path of samplesheet with SRA code or location of fastq files'

}

In case you started from SRA codes you also need to declare it in the params_mirna.yml

params{
  sra_mirna = true
}

The general command to run the pipeline is:

nextflow run multiomicsintegrator -params-file multiomicsintegrator/params_mirna.yml -profile docker

This will launch the pipeline with the docker configuration profile. See below for more information about profiles.

Note that the pipeline will create the following files in your working directory:

work                'Directory containing the nextflow working files'
<OUTDIR>            ' Location of where you want your results (defined by outdir)'
.nextflow_log       # Log file from Nextflow
# Other nextflow hidden files, e.g., history of pipeline runs and old logs.

Functionality

The pipeline initially downloads SRA codes and converts the runs into fastq files. Alternatively, you can provide local fastq files. It then performs quality control with FASTQC and then automatically detects and removes adapters with Trimgalore.

Each of the above steps can be skipped, for example if you don’t want to perform quality control, you can specify in the params_mirna.yml file:

params{
  skip_qc_mirna= true
}

It then employs salmon in order to obtain quantification files that are outputted in

/OUTDIR/salmon_mirna/sampleID/quant.sf

directory.

If you have performed the alignment step outside you can organise your data in the aforementioned way and specify the directory that holds the quant.sf files in the params_mirna.yml:

params{
  salmonDirmiRNA = '/path/to/directory_that_holds_quantification_files'
}

Note: All files need to be in the format:

salmonDirmiRNA :
- sampleID/
  - quant.sf

The sampleID is the same of the sampleID of the phenotype file (or the sample names of the samplesheet.csv)

If you want to skip the alignment step you need to specify the location of the count matrix and the respective phenotype (samplesheet_mirna.csv) you have, in the params_mirna.yml file:

params{
  count_matrix_mirna = 'path where count matrix is located'
  input_mirna = 'path where your phenotype file is located'
}

Preprocess

After the formation of the count matrix there is an optional module preprocess_matrix that performs preprocessing steps on the count matrix. Namely, the user can perform filtering, normalization and batch effect correction, depending on the state of their data.

Input_mirna should have a column named condition describing the states of the experiment (ctr vs treat) and one called “batch” describing batches of the experiment (if there is no batch then the replicate column is the batch). If the user wants other names they have to specify in the params_mirna.yml the column name of their conditions and that column name to be present in the input_mirna file:

params{
    mom_filt_method_mirna           = "filterByExp"  # filterByExp or choose a cutoff value
    mom_norm_method_mirna           = "quantile"     # calcNorm quantile
    mom_norm_condition_mirna           = "condition"   # must be column in samples info
    mom_norm_treatment_mirna           = "condition"   # must be column in samples info
    mom_batch_method_mirna          = "com" # com for combat, sva,  comsva for combat & sva, svacom for sva and comba, none
    mom_batch_condition_mirna       = "condition"    # which is the condition of interest, must be present in columns of sample info
    mom_batch_batch_mirna           = "batch"
}

DEA

At this stage, it is time to perform differential expression analysis. We provide three different algorithms for that, which we describe below.

Note

You need to specify which algorithm you are going to use in params_mirna.yml

params{
  alg_mirna     = 'edger' # Default
}

edgeR

params{
    dgergroupingfactor_mirna        =  "condition" # column name where your treatments are located
    edgerformulamodelmatrix_mirna   =  "~0 + condition" # design matrix, values have to be column names in the samplesheet_mirna.csv
    edgercontrasts_mirna            = "TNBC-non_TNBC"  # contrasts of interest. Values have to be present in the samplesheet_mirna.csv
}

DESeq2

Important note

For DESeq2 to run, the column of the treatments in the samplesheet_mirna.csv has to be named condition and the batches batch

params{
    batchdeseq2_mirna               = false # perform batch effect correction
    deseqFormula_mirna              = "~0 + condition"  # design matrix, values have to be column names in the samplesheet_mirna.csv
    con1_mirna                     = "mkc"   # control, has to be cell in samplesheet_mirna.csv
    con2_mirna                     = "dmso"  # treatment, has to be cell in samplesheet_mirna.csv
    deseq2single_matrix             = true   # if the input is a single matrix or a directory of files
}

rankprod

Inputs for to run RankProduct are the same, with a single difference: The condition column has to be named cl and the user has to asign 0 to controls and 1 to treatments

sampleID,cl
CONTROL_REP1,1
CONTROL_REP2,1
TREATMENT_REP1,0

Pathway Enrichment Analysis (PEA) —

The last step of the analysis is to perform pathway enrichment analysis with clusterprofiler or biotranslator

params{
    features                         = null # if you want to perform clusterprofiler as a standalone tool, specify directory of features here
    alg                        = "edger" # algoritmh you used to perform differential expression analysis or mcia
    genes_genespval                  = 1 # pval cutoff for genes
    mirna_genespval                  = 1 # pval cutoff for miRNA
    proteins_genespval               = 0.5 # pval cutoff for proteins
    lipids_genespval                 = 0.5 # pval cutoff for lipids
}

params{
    // BIOTRANSLATOR
    pea_mirna      = "biotranslator"
    biotrans_mirna_organism          = "hsapiens"
    biotrans_mirna_keytype          = "gene_symbol"
    biotrans_mirna_ontology         = "GO" // MGIMP, Reactome
}