Question 16 of 20 Intermediate Interpretation
How would you interpret a failed per base sequence content module in a FastQC report?
Short answer
It depends on the library. Random-hexamer RNA-seq and transposase libraries almost always fail on the first dozen or so bases because of priming bias, which does not harm expression estimates. A failure along the whole read points instead to overrepresented sequences, contamination or a biased library type.
Why
The module plots the share of A, C, G and T at each position in the read. In a random library the four lines should run roughly parallel. FastQC warns when the difference between A and T, or G and C, exceeds 10 percent at any position, and fails above 20 percent.
Its documentation explains that libraries primed with random hexamers, which includes nearly all RNA-seq, and those fragmented with transposases have a selection bias in roughly the first 12 bases. It states that this cannot be fixed by processing and does not seem to affect measuring expression. Other causes matter more.
Overrepresented sequences, such as adapter dimers or rRNA, distort composition and their sequence can be read off the plot. Some library types are biased by design. Bisulfite treatment converts most cytosines to thymines, so the plot fails although the library is normal.
Amplicon libraries, where every read starts at the same primer, also fail. So the reading depends on where the imbalance sits and what library made it.
A strong answer covers
- Warn above 10 percent and fail above 20 percent imbalance
- Early-position bias is normal for hexamer and transposase libraries
- Imbalance along the whole read suggests overrepresented sequences
- Bisulfite and amplicon libraries fail by design
- Interpret against the library type
Common mistakes
- Trimming the first bases of every RNA-seq read to make the module pass
- Rejecting a bisulfite library because the module failed
- Treating the FastQC pass or fail flag as a verdict on the sample
The interviewer might ask next
Which other FastQC modules would you check to confirm adapter dimers?