Alternative Splicing Analysis Using Nanopore Long-Reads RNA Sequencing Data

Description

This snakemake pipeline is for alternative splicing analysis using Nanopore long-read RNA sequencing data. The pipeline may be run on an HPC or in a local environment.

Major steps in this workflow include:

Basing calling using BONITO
Alignment using Minimap2
Splicing analysis using Stringtie2
Splicing analysis using Freddie

Software Requirements

User's guide

I. Input requirements

Edited config/config.yaml
Demultiplexed Nanopore fast5 data
Reference genome and annotation files

II. Editing the config.yaml

Basic parameters:

bonito: Path to the directory where Bonito is installed
raw: Path to the directory where demultiplexed fast5 data is stored. The name of sub-directory for each sample should be in this format: {sampleID}_fast5
out: Output directory, default without input: "output" in working directory
reference: Path to primary reference genome fasta file
gtf: Path to primary reference genome annotation gtf file
py: If use Pychopper to identify, orient and trim full-length Nanopore cDNA reads or not. Input "yes" if PCR amplification is included in the library preparation protocol; otherwise input "no".

Parameters for alignment quality filtering:

mq: Minmum mapping quality

Parameters for Stringtie2:

stringtie_opts: --conservative (conservative transcriptome assembly) --rf/--fr (stranded library), etc

Parameters for Freddie:

split: Path to the script freddie_split.py
segment: Path to the script freddie_segment.py
cluster: Path to the script freddie_cluster.py
isoforms: Path to the script freddie_isoforms.py
lic: Path to gurobi license
mr1: Minimum contrasting coverage support required for a breakpoint in Segment. Default: 3
mr2: Minimum read support allowed for an isoform in Cluster. Default: 3
mps: Maximum number of candidate breakpoints allowed per segmentation problem in Segment. Default: 50

Optional parameters to run the full analysis on HPV genomes:

(This function was originally designed for the HPV project, however, it can be applied to any secondary genome analysis of any species with references.)

hpv: Run the analysis on HPV genomes or not
hpv_ref: Path to HPV reference genome fasta file
hpv_gtf: Path to HPV reference genome annotation file

III. To run

Clone the repository to your working directory

git clone https://github.com/NCI-CGR/Nanopore_RNA-seq.git

Install required software
Create conda environment using the provided yaml file and activate it after installation:
```
conda env create -f env/nanopore_RNA.yaml
conda activate nanopore_RNA
```
Edit and save config/config.yaml
To run on an HPC using slurm job scheduler like NIH Biowulf (Pascal GPU or higher architecture required):

Edit config/cluster_config.yaml according to your HPC information and adjust the computing resources as needed. Run sbatch.sh to initiate pipeline running.
```
bash sbatch.sh
```

To run in a local environment:

export LD_LIBRARY_PATH={PATH}/conda/envs/nanopore_RNA/lib/:$LD_LIBRARY_PATH
export PYTHONPATH=$PYTHONPATH:{PATH}/conda/envs/nanopore_RNA/lib/python3.7/site-packages/
export GRB_LICENSE_FILE={PATH}/gurobi.lic
snakemake -p --cores 16 --keep-going --rerun-incomplete --jobs 300 --latency-wait 120 all

Look in log directory for logs for each rule

To view the snakemkae rule graph:

snakemake --rulegraph | dot -T png > nanopore_RNA.png

IV Output directory structure

{output directory}
├── alignment # Alignment on primary (eg human) genome
│   └── {sampleID}
├── alignment_hpv # Alignment on secondary (eg HPV) genome
│   └── {sampleID}
├── freddie # Splicing analysis on primary genome using Freddie
│   └── {sampleID}
├── freddie_hpv # Splicing analysis on secondary genome using Freddie
│   └── {sampleID}
├── reads # Base calling
│   └── {sampleID}
├── stringtie # Splicing analysis on primary genome using Stringtie2
│   └── {smapleID}
└── stringtie_hpv # Splicing analysis on secondary genome using Stringtie2
    └── {smapleID}

NCI-CGR / Nanopore_RNA-seq