CHO lncRNA annotation and differential expression analysis

Introduction

This respository enables the reproduction of the analysis described in:

Motheramgari et al. Expanding the Chinese hamster ovary cell long non-coding RNA transcriptome using RNASeq Biotechnology and Bioengineering 2020 https://doi.org/10.1002/bit.27467

Data availability:

NCBI Sequence Read Archive
European Nucleotide Archive

Dependancies

All the programmes must be added to the PATH to run the workflow

RNASeq data preprocesssing

Download the data from ENA

This is a simple way to dowload from ENA, for higher speed download use the Aspera client Total data download size: ~95G

mkdir -p data/ena
wget -q "ftp://ftp.sra.ebi.ac.uk/vol1/fastq/SRR105/057/SRR10572657/*" -P data/ena || { handle ; error ; }

wget -q "ftp://ftp.sra.ebi.ac.uk/vol1/fastq/SRR105/058/SRR10572658/*" -P data/ena || { handle ; error ; }

wget -q "ftp://ftp.sra.ebi.ac.uk/vol1/fastq/SRR105/059/SRR10572659/*" -P data/ena || { handle ; error ; }

wget -q "ftp://ftp.sra.ebi.ac.uk/vol1/fastq/SRR105/060/SRR10572660/*" -P data/ena || { handle ; error ; }

wget -q "ftp://ftp.sra.ebi.ac.uk/vol1/fastq/SRR105/061/SRR10572661/*" -P data/ena || { handle ; error ; }

wget -q "ftp://ftp.sra.ebi.ac.uk/vol1/fastq/SRR105/062/SRR10572662/*" -P data/ena || { handle ; error ; }

wget -q "ftp://ftp.sra.ebi.ac.uk/vol1/fastq/SRR105/063/SRR10572663/*" -P data/ena || { handle ; error ; }

wget -q "ftp://ftp.sra.ebi.ac.uk/vol1/fastq/SRR105/064/SRR10572664/*" -P data/ena || { handle ; error ; }

trim adapter sequences

Adapter trimming for the Illumina TruSeq adapters

mkdir data/cutadapt
cat data/sample_info.txt | cut -f 2 | tail -n 8 | while read sample; do
./scripts/trim_adapter.sh -s $sample  -i data/ena -o data/cutadapt
done

quality trimming

Trimmomatic for removing low quality bases and filtering resulting reads that are
too short. The script makes two subfolders in the trimmomatic folder for paired and unpaired reads

mkdir data/preprocessed
cat data/sample_info.txt | cut -f 2 | tail -n 8 | while read sample; do
./scripts/trim_quality.sh -s $sample -i data/cutadapt -o data/preprocessed -p 32
done

Read Mapping

Download the reference genome and NCBI annotation for CHO K1

. The sequence is also prepped for further analysis by retaining only the scaffold ID. In addition, a complementary annotation file is created from NCBI to help with annotation later

mkdir reference_genome
./scripts/prepare_genome.sh -v 98 -o reference_genome

make the STAR index

Use the reference genome to build an index for mapping the RNASeq reads

./scripts/make_star_index.sh -g reference_genome/ensembl_chok1_genome.fa -a reference_genome/ensembl_chok1_genome.gtf -p 32 -d reference_genome

map to CHOK1 genome

mkdir data/mapped
cat data/sample_info.txt | cut -f 2 | tail -n 8 | while read sample; do
./scripts/star_mapping.sh -s $sample -i data/preprocessed/paired -g reference_genome/star_index -o data/mapped -p 32
done

Genome guided assembly

string tie assembly

mkdir transcriptome_assembly
cat data/sample_info.txt | cut -f 2 | tail -n 8 | while read sample; do
  ./scripts/run_stringtie.sh -s $sample -i data/mapped -g reference_genome/ensembl_chok1_genome.gtf -o transcriptome_assembly -p 32
done

Merge individual stringtie assemblies and compare to ENSEMBL annotation

In this code the stringtie is separated into transcripts originating from annotated protein coding genes and non-coding RNAs

./scripts/stringtie_merge.sh -t transcriptome_assembly -g reference_genome/ensembl_chok1_genome.gtf -r reference_genome

lncRNA annotation

Make a directory to hold each of the different analyses

mkdir lncrna_annotation

Calculate the expression values for each transcript

The TPM expression value is calculated for each transcript assembled by StringTie

cat data/sample_info.txt | cut -f 2 | tail -n 8 | while read sample; do
  ./scripts/calc_tpm.sh -p 32 -s $sample -g transcriptome_assembly/stringtie.all.transcripts.gtf -o lncrna_annotation/TPM -b data/mapped
  done

List samples

cat data/sample_info.txt | cut -f 2 | tail -n 8 | while read sample; do
awk 'NR>1 {print $2}' data/sample_info.txt > lncrna_annotation/TPM/sample_list.txt
./scripts/tpm_matrix.sh -s lncrna_annotation/TPM/sample_list.txt -o lncrna_annotation/TPM
done

FEELNc analysis

The stringtie transcriptome assembly is used to predict lncRNAs using FEELNc

./scripts/run_feelnc.sh -G reference_genome/ensembl_chok1_genome.gtf -g transcriptome_assembly/non_protein_coding_stringtie.gtf -f reference_genome/ensembl_chok1_genome.fa -o lncrna_annotation/FEELnc

Assess FEELNc output using additional protein potential calculators, PFAM search and BLAST against protein and RNA databases

Use transdecoder to create a cDNA fasta file identify the longest ORF for each candidate lncRNA

./scripts/run_transdecoder.sh -g lncrna_annotation/FEELnc/feelnc_codpot_out/candidate_lncRNA.gtf.lncRNA.gtf -f reference_genome/ensembl_chok1_genome.fa -o lncrna_annotation/TRANSDECODER

CPAT coding prediction for FEELNc candidate lncRNAs

Determine protein coding potential using CPAT

./scripts/run_cpat.sh -f lncrna_annotation/TRANSDECODER/candidate_lncrna_cdna.fa -o lncrna_annotation/CPAT

CPC2 coding prediction for FEELnc candidate lncRNAs

Determine protein coding potential using CPC2

  ./scripts/run_cpc2.sh -f lncrna_annotation/TRANSDECODER/candidate_lncrna_cdna.fa -o lncrna_annotation/CPC2

Assess FEELnc candiate lncRNAs for the presence of protein domains

Search for known PFAM domains

./scripts/run_hmmscan.sh -t 32 -e 1e-5 -p lncrna_annotation/TRANSDECODER/longest_orfs.pep -o lncrna_annotation/PFAM

BLAST FEELnc candiate lncRNAs

Search SWISSPROT,RFAM & MIRBASE

./scripts/run_blast.sh  -t 32 -e 1e-5 -s lncrna_annotation/SWISSPROT -p lncrna_annotation/TRANSDECODER/longest_orfs.pep -m lncrna_annotation/MIRBASE -r lncrna_annotation/RFAM -n lncrna_annotation/TRANSDECODER/candidate_lncrna_cdna.fa

First stage of filtering

Remove candidates from FEELNc that don’t meet criteria

/usr/bin/Rscript R/filter_lncrna.R \
  "lncrna_annotation/FEELnc/feelnc_codpot_out/candidate_lncRNA.gtf.lncRNA.gtf" \
  "lncrna_annotation/CPC2/CPC2.analysis.txt" \
  "lncrna_annotation/CPAT/CPAT.analysis.txt" \
  "lncrna_annotation/SWISSPROT/blastp.outfmt6" \
  "lncrna_annotation/MIRBASE/blastn.outfmt6" \
  "lncrna_annotation/PFAM/pfam_domain_transcripts" \
  "lncrna_annotation/RFAM/blastn.outfmt6" \
  "lncrna_annotation/TPM/transcript_tpm_all_samples.tsv" \
  "lncrna_annotation/FEELnc/lncRNA_classes.txt" \
  "transcriptome_assembly/non_protein_coding_stringtie.gtf" \
  "lncrna_annotation"

Second stage of filtering

Here we further filter potential lncRNAs by removing monoexonic transcripts

determine synteny with human and mouse gencode lncRNAs

./scripts/orthology_analysis.sh -s lncrna_annotation/firstpass_filter/lncRNA.fasta -o lncrna_annotation

Filter monoexonic lncRNAs

keep only monoexonic lncRNAs that are:

1) antisense to a protein coding gene 2) are orthologous with human or mouse 3) annotated as lncRNA in ensembl

./scripts/filter_monoexonic_lncrnas.sh \
-o lncrna_annotation \
-g reference_genome/ensembl_chok1_protein.gtf \
-l lncrna_annotation/firstpass_filter/all_lncrna.gtf \
-k reference_genome/ensembl_lncrna_transcript.txt \
-a transcriptome_assembly/non_protein_coding_stringtie.gtf

Classify lncRNA types

/usr/bin/Rscript R/Rscript R/simplify_class.R \
  "lncrna_annotation/classification/second_classification.txt" \
  "lncrna_annotation/classification/final_classification.txt"

Differential expression analysis

Determine the lncRNAs that are differentially expressed between the temperature shift and non-temperature shift conditions

Gene level counting

mkdir differential_expression
cat data/sample_info.txt | cut -f 2 | tail -n 8 | while read sample; do
./scripts/htseq_count.sh -s $sample -m data/mapped -g transcriptome_assembly/stringtie.all.transcripts.gtf -o differential_expression/counts&
done

DESeq2 differential expression

mkdir differential_expression/results
Rscript R/run_deseq.R "differential_expression/counts" "differential_expression/results"