The first computational stage of an RNA-Seq workflow begins with raw sequencing reads.
Before quantification, normalization, or differential expression analysis, it is important to evaluate the quality of the sequencing data. Poor-quality reads can introduce technical artifacts that affect downstream results and interpretation.
Quality control (QC) helps determine whether sequencing data are suitable for analysis and whether corrective actions may be needed.
flowchart TD
A[Sequencing]
subgraph DP["Data Processing"]
B[Raw Reads]
C[Read Quality Control]
D[Read Processing & Quantification]
E[Count Matrix]
end
A --> B
B --> C --> D --> Eflowchart TD
A[Sequencing]
subgraph DP["Data Processing"]
B[Raw Reads]
C[Read Quality Control]
D[Read Processing & Quantification]
E[Count Matrix]
end
A --> B
B --> C --> D --> E
This chapter focuses on the transition from raw reads to quality-assessed sequencing data.
RNA-Seq sequencing typically produces FASTQ files.
A FASTQ file contains:
Each base is assigned a Phred quality score that reflects the probability of an incorrect base call.
Higher quality scores indicate greater confidence in the sequencing result.
Quality control aims to answer questions such as:
These assessments help determine whether downstream analyses can proceed confidently.
Typical RNA-Seq QC metrics include:
No single metric determines success or failure. Interpretation requires considering multiple metrics together.
Per-base quality plots show how sequencing quality changes across read positions.
A common pattern is:
Large quality drops may indicate potential problems.
The goal is not perfection but confidence that sequencing quality supports downstream analyses.
Quality scores are commonly represented as Phred scores.
A simplified interpretation is:
| Phred Score | Error Probability |
|---|---|
| 10 | 1 in 10 |
| 20 | 1 in 100 |
| 30 | 1 in 1,000 |
| 40 | 1 in 10,000 |
Higher scores indicate lower expected sequencing error rates.
Adapter sequences may appear when sequencing reads extend beyond the biological insert.
Common signs include:
Depending on the workflow, adapters may need to be removed before downstream processing.
GC-content distributions can help identify unusual sequencing patterns.
Unexpected GC-content may reflect:
Interpretation should consider the biological context of the study.
Duplicate reads are not always problematic in RNA-Seq.
Highly expressed genes naturally generate many similar reads.
However, extremely high duplication rates may sometimes indicate:
Duplication metrics should be interpreted carefully.
RNA-Seq samples should have sufficient sequencing depth to address the study question.
Typical considerations include:
Large differences in sequencing depth may influence downstream analyses and should be investigated.
A commonly used RNA-Seq QC tool is FastQC.
FastQC summarizes multiple quality metrics, including:
FastQC reports provide a useful starting point for evaluating sequencing quality.
Example command:
fastqc sample.fastq.gzWhen many samples are analyzed, MultiQC can aggregate QC reports into a single summary.
Example command:
multiqc .MultiQC makes it easier to identify trends and outliers across samples.
A typical QC workflow may look like:
FASTQ Files
↓
FastQC
↓
Review Reports
↓
Identify Issues
↓
MultiQC Summary
↓
Proceed to QuantificationThe goal is to understand data quality before generating expression measurements.
Quality control is primarily an assessment process.
A warning does not automatically mean:
QC findings should be interpreted in the context of:
Context matters.
Common QC mistakes include:
QC is a process of evidence-based evaluation rather than rule-based filtering.
Before moving to quantification, confirm that:
Raw read quality control is the first computational checkpoint in the RNA-Seq system.
Its purpose is to understand the quality of the sequencing data, identify potential technical issues, and provide confidence that downstream analyses are based on reliable input data.
The next chapter focuses on read processing and quantification, where sequencing reads are converted into quantitative expression measurements.