GSE77633 Processing Pipeline

Publication

Nature methods (2016) — PMID 27018577

Dataset

Enhanced CLIP (eCLIP) enables robust and scalable transcriptome-wide discovery and characterization of RNA binding protein binding sites [iCLIP]

1. 1

   Sequencing reads from CLIP-seq libraries were first trimmed of polyA tails, adapters, and low quality ends using cutadapt with parameters --match-read-wildcards --times 2 -e 0 -O 5 --quality-cutoff' 6 -m 18 -b TCGTATGCCGTCTTCTGCTTG -b ATCTCGTATGCCGTCTTCTGCTTG -b CGACAGGTTCAGAGTTCTACAGTCCGACGATC -b TGGAATTCTCGGGTGCCAAGG -b AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA -b TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT.

   cutadapt v1.18 GitHub

   $ Bash example

   # Install cutadapt (if not already installed)
   # conda install -c bioconda cutadapt
   
   # Define input and output filenames (placeholders)
   # Replace 'input.fastq.gz' with your actual input FASTQ file
   # Replace 'output.fastq.gz' with your desired output FASTQ file
   INPUT_FASTQ="input.fastq.gz"
   OUTPUT_FASTQ="output.fastq.gz"
   
   cutadapt \
       --match-read-wildcards \
       --times 2 \
       -e 0 \
       -O 5 \
       --quality-cutoff 6 \
       -m 18 \
       -b TCGTATGCCGTCTTCTGCTTG \
       -b ATCTCGTATGCCGTCTTCTGCTTG \
       -b CGACAGGTTCAGAGTTCTACAGTCCGACGATC \
       -b TGGAATTCTCGGGTGCCAAGG \
       -b AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA \
       -b TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT \
       -o "${OUTPUT_FASTQ}" \
       "${INPUT_FASTQ}"

   Copy View on GitHub
2. 2

   Reads were then mapped against a database of repetitive elements derived from RepBase18.05.

   bowtie2 (Inferred with models/gemini-2.5-flash) v2.4.5 (Inferred with models/gemini-2.5-flash) GitHub

   $ Bash example

   # Install bowtie2 if not already installed
   # conda install -c bioconda bowtie2
   
   # Define input reads (adjust for single-end or paired-end as needed)
   # For paired-end reads:
   READS_R1="reads_R1.fastq.gz"
   READS_R2="reads_R2.fastq.gz"
   
   # For single-end reads (uncomment and adjust if applicable):
   # READS_SINGLE="reads.fastq.gz"
   
   OUTPUT_SAM="mapped_to_repeats.sam"
   UNMAPPED_PREFIX="unmapped_to_repeats" # Prefix for unmapped reads files (e.g., unmapped_to_repeats_1.fastq.gz, unmapped_to_repeats_2.fastq.gz)
   
   # Define repetitive elements database (RepBase18.05)
   # RepBase is a commercial database. Users typically obtain a license or use derived public datasets.
   # For demonstration, assume 'RepBase18.05.fasta' is available in the working directory.
   REPEATS_FASTA="RepBase18.05.fasta"
   REPEATS_INDEX_PREFIX="RepBase18.05_index"
   
   # Build Bowtie2 index for repetitive elements
   # This step only needs to be run once per reference database
   # bowtie2-build "${REPEATS_FASTA}" "${REPEATS_INDEX_PREFIX}"
   
   # Map reads against the repetitive elements database (Paired-end example)
   # Using --very-sensitive-local for robust mapping to potentially fragmented repeats
   # --un-conc-gz to output unmapped reads (paired-end) for subsequent mapping to the main genome
   bowtie2 \
     --very-sensitive-local \
     -x "${REPEATS_INDEX_PREFIX}" \
     -1 "${READS_R1}" \
     -2 "${READS_R2}" \
     --un-conc-gz "${UNMAPPED_PREFIX}" \
     -S "${OUTPUT_SAM}"
   
   # If using single-end reads, the command would be:
   # bowtie2 \
   #   --very-sensitive-local \
   #   -x "${REPEATS_INDEX_PREFIX}" \
   #   -U "${READS_SINGLE}" \
   #   --un-gz "${UNMAPPED_PREFIX}.fastq.gz" \
   #   -S "${OUTPUT_SAM}"
   
   # Optional: Convert SAM to BAM and sort
   # samtools view -bS "${OUTPUT_SAM}" | samtools sort -o "${OUTPUT_SAM%.sam}.bam"
   # samtools index "${OUTPUT_SAM%.sam}.bam"

   Copy View on GitHub
3. 3

   Bowtie version 1.0.0 with parameters -S -q -p 16 -e 100 -l 20 was used to align reads against an index generated from Repbase sequences (Langmead et al., 2009).

   Bowtie v1.0.0 GitHub

   $ Bash example

   # Install Bowtie (version 1.0.0 might require specific steps or older environments)
   # For example, if using conda:
   # conda create -n bowtie1_env bowtie=1.0.0
   # conda activate bowtie1_env
   
   # Align reads using Bowtie 1.0.0
   # Assuming 'repbase_index' is the prefix for the Bowtie index files (e.g., repbase_index.1.ebwt, repbase_index.2.ebwt, etc.)
   # and 'reads.fastq' contains the input reads.
   # The output will be in SAM format due to the -S flag.
   bowtie -S -q -p 16 -e 100 -l 20 repbase_index reads.fastq > aligned_reads.sam

   Copy View on GitHub
4. 4

   Reads not mapped to Repbase sequences were aligned to the hg19 human genome (UCSC assembly) using STAR (Dobin et al., 2013) version 2.3.0e with parameters --outSAMunmapped Within –outFilterMultimapNmax 1 –outFilterMultimapScoreRange 1.

   STAR v2.3.0e

   $ Bash example

   # Install STAR (example using conda)
   # conda install -c bioconda star=2.3.0e
   
   # Define variables (replace with actual paths)
   GENOME_DIR="/path/to/hg19_STAR_index"
   INPUT_FASTQ="unmapped_reads.fastq"
   OUTPUT_PREFIX="aligned_to_hg19_"
   
   # Run STAR alignment
   STAR \
     --genomeDir "${GENOME_DIR}" \
     --readFilesIn "${INPUT_FASTQ}" \
     --outSAMtype BAM SortedByCoordinate \
     --outFileNamePrefix "${OUTPUT_PREFIX}" \
     --outSAMunmapped Within \
     --outFilterMultimapNmax 1 \
     --outFilterMultimapScoreRange 1

   Copy
5. 5

   Reads that were PCR replicates were removed from each CLIP-seq library using a custom script.

   CLIP-seq v1.19 GitHub

   $ Bash example

   # Install samtools if not already installed
   # conda install -c bioconda samtools
   
   # Define input and output file names
   INPUT_BAM="aligned_reads.bam"
   OUTPUT_DEDUP_BAM="deduplicated_reads.bam"
   TEMP_FIXMATE_BAM="temp_fixmate.bam"
   TEMP_NAMESORT_BAM="temp_namesort.bam"
   FINAL_COORD_SORTED_BAM="deduplicated_reads_coord_sorted.bam"
   
   # 1. Add mate score tags and fixmate information
   # This step is crucial for samtools markdup to correctly identify paired-end duplicates.
   # The output is unsorted.
   samtools fixmate -m "${INPUT_BAM}" "${TEMP_FIXMATE_BAM}"
   
   # 2. Sort by read name
   # samtools markdup requires name-sorted input for optimal performance.
   samtools sort -n "${TEMP_FIXMATE_BAM}" -o "${TEMP_NAMESORT_BAM}"
   
   # 3. Remove PCR duplicates
   # -r: Remove duplicate reads (instead of just marking them).
   # -s: Output statistics to stderr (optional, but good for logging).
   samtools markdup -r "${TEMP_NAMESORT_BAM}" "${OUTPUT_DEDUP_BAM}" 2> "${OUTPUT_DEDUP_BAM}.stats"
   
   # 4. Re-sort the deduplicated BAM by coordinate
   # This is often required for downstream tools like peak callers or visualization.
   samtools sort -o "${FINAL_COORD_SORTED_BAM}" "${OUTPUT_DEDUP_BAM}"
   
   # 5. Index the final BAM file
   # Indexing allows for fast retrieval of reads by genomic location.
   samtools index "${FINAL_COORD_SORTED_BAM}"
   
   # Clean up temporary files
   rm "${TEMP_FIXMATE_BAM}" "${TEMP_NAMESORT_BAM}" "${OUTPUT_DEDUP_BAM}"

   Copy View on GitHub
6. 6

   Briefly one read with a unique barcode was kept at each nucleotide position when more than one with the same barcode was mapped to the same location

   umi_tools dedup (Inferred with models/gemini-2.5-flash) v1.1.2 (Inferred with models/gemini-2.5-flash) GitHub

   $ Bash example

   # Install umi_tools if not already installed
   # conda install -c bioconda umi_tools
   
   # Placeholder for input and output files
   INPUT_BAM="aligned_reads_with_umis.bam"
   OUTPUT_BAM="deduplicated_reads.bam"
   
   # Deduplicate reads based on Unique Molecular Identifier (UMI) and mapping position.
   # This command assumes that the UMI is already stored in a BAM tag named 'UB' (UMI Barcode).
   # If the UMI is part of the read name, you would typically use '--extract-method'
   # and '--umi-separator' or '--umi-pattern' instead.
   # For example, if UMI is at the start of the read name separated by '_':
   # umi_tools dedup -I "${INPUT_BAM}" -S "${OUTPUT_BAM}" --extract-method=string --umi-separator='_'
   # The default behavior of umi_tools dedup is to group reads by mapping position
   # and UMI, then keep one representative read.
   umi_tools dedup \
       -I "${INPUT_BAM}" \
       -S "${OUTPUT_BAM}" \
       --umi-tag=UB \
       --log "${OUTPUT_BAM%.bam}.log"

   Copy View on GitHub
7. 7

   Clusters were then assigned using the CLIPper software with parameters --bonferroni --superlocal --threshold- software (Lovci et al., 2013).

   CLIPper v1.0 (Inferred from Lovci et al., 2013) GitHub

   $ Bash example

   # Install CLIPper (assuming Python environment)
   # git clone https://github.com/yeolab/clipper.git
   # cd clipper
   # python setup.py install # Or ensure clipper.py is in your PATH or run directly
   
   # Example usage of CLIPper for peak calling
   # Replace 'path/to/genome.sizes' with the actual path to your genome size file (e.g., hg38.chrom.sizes)
   # Replace 'treatment.bam' and 'control.bam' with your actual aligned BAM files
   # The '--threshold-' parameter in the description was incomplete. Assuming it should be '--threshold <value>'.
   python clipper.py \
       -s path/to/genome.sizes \
       -o clipper_output_prefix \
       treatment.bam \
       control.bam \
       --bonferroni \
       --superlocal \
       --threshold 0.05 # Placeholder value for threshold, as it was incomplete in description

   Copy View on GitHub

Tools Used