LoRID is a versatile pipeline to assemble and describe isoforms from long read sequencing using a new annotation tool called SOSTAR: iSofOrmS annoTAtoR. The pipeline is divided into two modules that can be run separately.

• First module: The first module performs alignment (in two rounds) using minimap2, assembled isoforms and computed isoform expression using StringTie.
• Second module: The second module provides a descriptive and comprehensive annotation of each assembled isoform. Isoforms are described relative to reference transcripts (provided by user) by an annotation including only alternative splicing events.

This pipeline is a Snakemake workflow.

Dependencies:

• Snakemake : 7.32.4 or greater.
• Singularity : 3.5.3 or greater.

The workflow automatically uses a docker image which contains the other tools required.

Download the Snakefile and rules of LoRID pipeline:

git clone https://github.com/LBGC-CFB/LoRID.git
cd ./LoRID

To configure this workflow, modify the ./config/config.yaml file according to your needs.

• genome: Fasta file of reference genome. Example in ./tests/hg19_chr17.fa.

• ensembl_annot: GTF annotation file of known reference transcrits described in databases (Ensembl, Refseq, ...). Example in ./tests/gencode.v19.annotation_chr17.gtf.

• transcripts_list: Txt file of the reference transcripts used for the annotation step. Name must match the "transcript_id" attribute from the ensembl_annot file. Example in ./tests/reference_transcripts_chr17.txt.

• indir: Path of the directory containing the .fastq.gz input files.

• outdir: Path of the outdirectory.

• samples: names of the differents samples.

• threads: number of threads to use.

• bedtools: option to indicate whether or not bedtools will be used to filter out transcripts from the genes specified in the reference transcript file. "True" or "False".

Given that the workflow has been properly configured, it can be executed as follows:

cd ./workflow
snakemake --use-singularity --cores

LoRID/outdir
├── alignment
│   └── aligned
│       ├── {sample}.aligned.bam
|       ├── {sample}.aligned.bam.bai
│       └── ...
│   └── realigned
│       ├── {sample}.realigned.bam
|       ├── {sample}.realigned.bam.bai
│       └── ...
├── assembly
│   └── aligned
│       ├── {sample}.assembly.aligned.gtf
│       └── ...
│   └── realigned
│       ├── {sample}.assembly.realigned.gtf
│       └── ...
│   ├── transcripts.merged.aligned.filter.bed
│   ├── transcripts.merged.aligned.filter.gtf
│   └── transcripts.merged.realigned.filter.gtf
├── expression
│   ├── {sample}.expression.gtf
│   └── ...
├── ref_transcripts_annotation.bed
├── ref_transcripts_annotation.gtf
└── SOSTAR_annotation_table_results.xlsx

• alignment folder: contains subfolders of all aligned and sorted {sample} <.bam> with their corresponding index <.bai> either on the first (aligned subfolder) or the second (realigned subfolder) round of alignment.
• assembly folder: contains subfolders of all assembled <.gtf> {sample} with StringTie either on the first (aligned subfolder) or the second (realigned subfolder) round of alignment plus the global <.gtf> and it corresponding <.bed> file of the merge isoforms.
• expression folder: contains all assembled <.gtf> with expression metrics computed by StringTie.
• ref_transcripts_annotation.bed: <.bed> file containing the coordinates of the genes specified by the user in the reference transcript file. Used for the bedtools intersect option to filter transcripts for these genes.
• ref_transcripts_annotation.gtf: <.gtf> annotation file filtered out of reference transcrits specified by the user from ensembl annot.
• SOSTAR_annotation_table_results.xlsx: final output file containing descriptive annotation and expression metrics of each transcript in the cohort (see SOSTAR annotation output file section for more informations).

The SOSTAR module can be used as a stand-alone tool, as it is a simple Python script. Using <.gtf> files of assembled isoforms from any assembly method, it provides a descriptive and comprehensive annotation of each assembled isoform. This module is compatible with any gtf file using the tag attribute gene_name that match to the name of reference transcript feature provided by user.

python3 scripts/SOSTAR.py -I {INPUT} -O {OUTPUT} -R {REF_GTF_COORD}

Options:

• -I, --input : /Path/to/gtf files

• -O, --output : /Path/to/output directory

• -R, --ref_gtf_coord : gtf annotation file filtered on reference transcripts used for the annotation

SOSTAR generates a spreadsheet file in <.xlsx> format:

• transcript_id : transcript identifier
• chr : chromosome of transcript
• start : start position of the transcript
• end : end position of the transcript
• strand : strand of the transcript
• gene : gene associated to the transcript
• annot_ref : exon range of the reference transcript
• annot_find : SOSTAR annotation of transcript (see section SOSTAR nomenclature for more details).
• sample : transcript coverage
• occurence : number of transcript occurrences in cohort

Isoforms are described relative to reference transcripts (provided by user) by an annotation including only the alternative splicing events. SOSTAR uses the systematic exon numbering. Some conventions were established to annotate the alternative splicing events:

Nomenclature example: Black boxes: exon, black lines: intron, red boxes: exon (or part of exon) skipping, green boxes: novel exon (or part of exon).

Camille AUCOUTURIER @AUCAM

This project is under GPL-3.0 License, see LICENCE for more details.