Lesson 2: Precision Quality Control (QC)

Why quality control matters

Before downstream analysis, assess raw read quality to detect artifacts that can bias results.

Common issues include:

• Adapter contamination
  • Synthetic adapter sequence is retained in reads
• Base-calling errors
  • Quality often drops toward read ends in long or noisy runs

Phred quality score

Phred score (Q) expresses the probability (P) that a base call is incorrect.

Q = -10 * log10(P)

Reference points:

Q10  -> error probability 1 in 10     -> 90% accuracy
Q20  -> error probability 1 in 100    -> 99% accuracy
Q30  -> error probability 1 in 1,000  -> 99.9% accuracy

Tools: FastQC and MultiQC

Use two standard tools together:

• FastQC: generates per-sample quality reports
• MultiQC: aggregates many FastQC outputs into one comparison report

Hands-on workflow

1) Activate environment

conda activate bioinfo

2) Run FastQC on training reads

mkdir -p ~/Training/QC_Results
fastqc ~/Training/short_reads/unpaired/*.fastq.gz -o ~/Training/QC_Results

3) Aggregate reports with MultiQC

cd ~/Training/QC_Results
multiqc .

What to inspect in the MultiQC report

Focus on these sections:

1. Per-base sequence quality
2. Adapter content
3. Overrepresented sequences

If quality tails drop below about Q20 or adapter content rises, trimming should be included before assembly.