11-miRNA.Rmd

# miRNA

The `miRDeep2` is one of the most popular tools for discovering known
and novel miRNAs from small RNA sequencing data. We have wrapped the
mapping and quantification steps into an `Rcwl` pipeline which is
ready to load and use. More details about `miRDeep2` can be found
here: <https://github.com/rajewsky-lab/mirdeep2>.

```{r}
miRDeep2PL <- cwlLoad("pl_miRDeep2PL")
plotCWL(miRDeep2PL)
```

Here We also use the data from the above GitHub repository as an example.
<https://github.com/rajewsky-lab/mirdeep2/tree/master/tutorial_dir>

```{r, eval=FALSE}
git2r::clone("https://github.com/rajewsky-lab/mirdeep2", "data/miRNA")
list.files("data/miRNA/tutorial_dir")
```

### Reference index

First, we need to build indexes for the miRNA reference with the
`Rcwl` tool `bowtie_build`. This is only required to be performed once
for each refernce genome.

```{r}
bowtie_build <- cwlLoad("tl_bowtie_build")
inputs(bowtie_build)
```

```{r, eval=FALSE}
bowtie_build$ref <- "data/miRNA/tutorial_dir/cel_cluster.fa"
bowtie_build$outPrefix <- "cel_cluster"
idxRes <- runCWL(bowtie_build, outdir = "output/miRNA/genome",
                 showLog = TRUE, logdir = "output/miRNA")
```

Then the indexed reference files are generated in the output directory
defined in `outdir`.

```{r}
file.copy("data/miRNA/tutorial_dir/cel_cluster.fa",
          "output/miRNA/genome/cel_cluster.fa")
dir("output/miRNA/genome")
```

### Run miRDeep2 pipeline

To run the pipeline for all samples parallelly, we need to prepare the
inputs for arguments of `inputList` and `paramList`.

```{r}
inputs(miRDeep2PL)
```

To mimic multiple samples, here we just repeat to use the input reads
as if they are two different samples.

```{r}
reads <- list(sample1 = "data/miRNA/tutorial_dir/reads.fa",
              sample2 = "data/miRNA/tutorial_dir/reads.fa")

inputList <- list(reads = reads)
paramList <- list(adapter = "TCGTATGCCGTCTTCTGCTTGT",
                  genome = "output/miRNA/genome/cel_cluster.fa",
                  miRef = "data/miRNA/tutorial_dir/mature_ref_this_species.fa",
                  miOther = "data/miRNA/tutorial_dir/mature_ref_other_species.fa",
                  precursors = "data/miRNA/tutorial_dir/precursors_ref_this_species.fa",
                  species = "C.elegans")
```

Let's run the pipeline with two computing nodes.

```{r, eval=FALSE}
mirRes <- runCWLBatch(miRDeep2PL, outdir = "output/miRNA",
                      inputList, paramList,
                      BPPARAM = BatchtoolsParam(
                          workers = 2, cluster = "multicore"))
```

The results are collected in the output directory defined in the `outdir`. 

```{r}
dir("output/miRNA/sample1")
```