lncRNA Analysis Pipeline

LncRNA framework for identification of lncRNA dysregulations from RNAseq data.

Overview

The lncRNA framework pipeline provides a comprehensive solution for lncRNA analysis, encompassing preprocessing, alignment, transcript quantification, differential expression analysis, and functional annotation. This document outlines the steps for processing a single sample. For batch processing, refer to the scripts folder.

Prerequisites

Required Tools

1. FastQC
2. fastp
3. STAR
   • Use --quantMode GeneCounts for quantification.
4. SAMtools
5. StringTie
   • Use -e -B for quantification.
   • Use StringTie Merge with -G for reference-based merging.
6. gffcompare (use -G for compatibility).
7. awk (for processing files).
8. CPAT (for coding potential assessment).
9. featureCounts (use -p -B -C for paired-end data).
10. DESeq2 (for differential expression analysis).
11. FEELnc (for functional annotation).
12. Miranda (for target prediction).
13. ClusterProfiler (for pathway enrichment).
14. Visualization tools:
    • ggplot2
    • igraph
15. GENCODE Annotations

Required Reference

• Genome: GRCh38.p14
• Annotation Files: GENCODE v47 GTF/GFF

---

Pipeline Steps

Step 1: Download Data from SRA

fastq-dump SRR24709142 --split-3 --gzip

Step 2: Quality Control of Raw Reads

fastqc SRR24709142_1.fastq.gz SRR24709142_2.fastq.gz -o qc_results

Step 3: Trimming and Filtering

fastp -i SRR24709142_1.fastq.gz -I SRR24709142_2.fastq.gz -o SRR24709142_1_clean.fastq.gz -O SRR24709142_2_clean.fastq.gz -h SRR24709142_report.html --trim_poly_g --cut_tail --cut_mean_quality 20 --length_required 50

Step 4: Generate STAR Index

STAR --runThreadN 50 --runMode genomeGenerate --genomeDir human_genome/STAR_index --genomeFastaFiles human_genome/GRCh38.p14.genome.fa --sjdbGTFfile human_genome/gencode.v47.long_noncoding_RNAs.gtf

Step 5: Align Reads to Reference Genome

STAR --runThreadN 50 --genomeDir human_genome/STAR_index --readFilesIn SRR24709142_1_clean.fastq.gz SRR24709142_2_clean.fastq.gz --readFilesCommand zcat --outSAMtype BAM SortedByCoordinate --outFileNamePrefix SRR24709142_ --quantMode TranscriptomeSAM GeneCounts --outSAMstrandField intronMotif

Step 6: Assemble Transcripts

cufflinks -g human_genome/gencode.v47.long_noncoding_RNAs.gtf -o SRR24709142_cufflinks_out -p 8 SRR24709142_Aligned.sortedByCoord.out.bam

Step 7: Create Transcript Merge List

Create a mergelist.txt file with paths to all transcripts.gtf files.

Step 7a: Merge GTF

cat $(<mergelist.txt)> merged.gtf

Step 8: Merge Transcripts

Step 8a: Remove duplicate coordinate-based entries

awk '!seen[$1, $4, $5]++' merged.gtf > merged_without_duplicates.gtf

Step 8b: Remove duplicate transcript ID-based entries

python Remove_Duplicate_transcript_ID.py merged_without_duplicates.gtf output_merged_without_duplicates.gtf

Step 9: Remove Mitochondrial Chromosome

awk '/^>/{p=($0!="chrM MT")} p' human_genome/GRCh38.p14.genome.fa > human_genome/GRCh38.p14.genome_no_chrM.fa

Step 10: Extract lncRNA sequences from genome fasta

gffread -w filtered_lncRNAs.fa -g human_genome/GRCh38.p14.genome_no_chrM.fa merged_without_duplicates.gtf

Step 11: Assess Coding Potential

Download pre-built models and run:

cpat.py -g filtered_lncRNAs.fa -x Human_Hexamer.tsv -d Human_logitModel.RData -o cpat_output.txt

Step 12: Cross-referencing with GTF file (filtered_gencode.gtf) to update annotations

The output file (cpat_output.txt) is filtered to retain only transcripts with low coding potential scores (i.e., non-coding). A typical threshold for non-coding classification is applied (e.g., coding potential score < 0.38 for human lncRNAs).

awk '$4 < 0.38' cpat_output.txt > high_confidence_lncRNAs.txt

This command filters the 4th column (coding potential score) and extracts only those classified as non-coding. The high-confidence lncRNAs are cross-referenced with GTF files (e.g., filtered_gencode.gtf) to update annotations, ensuring they align with transcript IDs of interest. The high-confidence lncRNAs identified in high_confidence_lncRNAs.txt are used as input for functional annotation (e.g., with FEELnc_filter.pl) to classify lncRNAs into various functional categories (e.g., intergenic or antisense).

Step 13: Extract transcript IDs from the filtered CPAT output

awk '{print $1}' high_confidence_lncRNAs.txt > high_confidence_transcripts.txt

Step 13_a: Cross-reference the filtered transcripts with the GTF file

grep -Ff high_confidence_transcripts.txt merged_without_duplicates.gtf > updated_filtered_gencode.gtf

Step 13_b: Verify the updated GTF file (optional)

head updated_filtered_gencode.gtf

Step 14: Quantify Transcript Abundance

featureCounts -a merged_without_duplicates.gtf -o SRR24709142_counts.txt -T 8 -p -B -C SRR24709142_Aligned.sortedByCoord.out.bam

Step 15: Process Count Files

Extract and combine processed data:

forfilein*_counts.txt;do
sed '1d'"$file"| awk -v OFS="\t"'{print $1, $7}'>"processed_$file"done
cut -f1 processed_SRR24709142_counts.txt > combined_file.txt
forfilein processed_*_counts.txt;do
awk '{print $2}'"$file"| paste combined_file.txt - > temp.txt
mv temp.txt combined_file.txt
done

Step 16: Differential Expression Analysis

Use R with DESeq2 for differential expression analysis and visualization. Refer to the R scripts in repository.

Step 17: Heatmap

Refer to the R scripts in repository.

Step 18: Functional Annotation

FEELnc_filter.pl -i updated_filtered_gencode.gtf -a gencode.v47.annotation.gtf -b transcript_biotype=protein_coding -l > candidate_lncRNA.gtf
awk 'BEGIN{FS="\t"; OFS="\t"} {match($9, /gene_id \"([^\"]+)\"/, a); if (a[1] != "") print a[1]}' candidate_lncRNA.gtf | sort | uniq > lncRNA_target_genes.txt

Step 19: Pathway Enrichment Analysis

Use clusterProfiler in R for pathway enrichment analysis. Refer to the R scripts in repository.

---

Output Files

1. Processed Counts: combined_file.txt
2. Differential Expression Results: differential_expression_results_with_labels.csv
3. Pathway Enrichment Results: pathway_enrichment_results_ENSEMBL_ID.csv

---

Notes

• Ensure all tools are installed and configured in a dedicated Conda environment.
• Use sufficient computational resources for STAR alignment and StringTie assembly.
• Verify input file formats before running each step.