3: Raw Read Quality Control & Adapter Trimming

Chapter 3: Raw Read Quality Control & Adapter Trimming

Raw reads off the sequencer are contaminated with residual adapter sequences, low-quality base calls at the 3' tails, sequencing artifacts, and optical flow cell anomalies. Passing unfiltered reads directly into aligners degrades mapping accuracy and causes false-positive variant calls. Quality control and pre-processing are non-negotiable first steps in every production NGS pipeline.

1. Deconstructing the FastQC Report

FastQC provides a rapid modular assessment of high-throughput sequence data:

• Per Base Sequence Quality: Boxplots depicting Q-score distribution per cycle. Scores above Q30 (green zone) indicate high fidelity. If median scores drop below Q20, trimming is necessary.
• Per Base Sequence Content: Evaluates base proportions (A, T, G, C) across cycles. While genomic DNA should exhibit equal lines (e.g. ~25% each depending on GC%), RNA-seq libraries often display severe nucleotide bias in the first 10–12 bases due to random hexamer priming during reverse transcription.
• Per Sequence GC Content: Compares the observed GC distribution to a theoretical normal curve. Bimodal distributions indicate sample contamination (e.g., bacterial infection or mycoplasma).
• Adapter Content: Identifies the proportion of reads containing Illumina universal adapter sequences, caused when library insert sizes are shorter than the read length (read-through).

2. High-Performance Pre-Processing with fastp

fastp is an ultra-fast, multithreaded preprocessor written in C++ that performs adapter trimming, quality filtering, poly-G clipping, and base correction in a single pass:

# Execute fastp on paired-end Illumina WGS reads
fastp \
    -i sample_R1.fastq.gz \
    -I sample_R2.fastq.gz \
    -o sample_trimmed_R1.fastq.gz \
    -O sample_trimmed_R2.fastq.gz \
    --detect_adapter_for_pe \
    --cut_front --cut_tail \
    --cut_window_size 4 --cut_mean_quality 20 \
    --qualified_quality_phred 20 \
    --unqualified_percent_limit 30 \
    --length_required 50 \
    --trim_poly_g \
    --thread 16 \
    --html sample_fastp_report.html \
    --json sample_fastp_report.json

3. Cohort Quality Aggregation with MultiQC

When sequencing hundreds of samples, inspecting individual FastQC or fastp reports is impossible. MultiQC recursively scans directories and aggregates all logs into a single interactive HTML dashboard:

# Run FastQC on all FASTQ files in parallel
fastqc -t 8 -o qc_reports/ raw_data/*.fastq.gz

# Aggregate FastQC, fastp, and aligner logs into a unified cohort dashboard
multiqc qc_reports/ -o cohort_qc_dashboard/