FASTQ Quality Assurance Tools

The first order of business after receiving sequencing data should be to check your data quality. This often-overlooked step helps guide the manner in which you process the data, and can prevent many headaches.

FastQC

FastQC is a tool that produces a quality analysis report on FASTQ files.

Useful links:

First and foremost, the FastQC "Summary" should generally be ignored. Its "grading scale" (green - good, yellow - warning, red - failed) incorporates assumptions for a particular kind of experiment, and is not applicable to most real-world data. Instead, look through the individual reports and evaluate them according to your experiment type.

The FastQC reports I find most useful are:

  1. The Per base sequence quality report, which can help you decide if sequence trimming is needed before alignment.

2. The Per Sequence Quality Score report, which can tell you if a subset of your reads just have poor quality scores. These reads can be completely filtered from analysis.

3. The Sequence Duplication Levels report, which helps you evaluate library enrichment / complexity. But note that different experiment types are expected to have vastly different duplication profiles.

 


DETOUR: What are PCR duplicates?

  1. The Overrepresented Sequences report, which helps look for dominant sequences.

    5. Adapter content report, which tells you about adapter contamination.



Note: For many of its reports, FastQC analyzes only the first 200,000 sequences in order to keep processing and memory requirements down.

Running FastQC

FastQC is available on lonestar5 as a module.

Here's how to run FastQC on our sample data:

Running FastQC example
module load biocontainers
module load fastqc
fastqc data/Sample1_R1.fastq

Exercise: FastQC results

What did FastQC create?

 Answer
ls -l shows something like this
drwxrwx--- 4 daras G-801020  32768 May 16 14:03 Sample1_R1_fastqc
-rw-rw---- 1 daras G-801020 186116 May 16 13:58 Sample1_R1_fastqc.zip

Looking at FastQC output

You can't run a web browser directly from your "dumb terminal" command line environment. The FastQC results have to be transferred to your computer to look at the html report.


Exercise: Should we trim this data?

Based on this FastQC output, should we trim this data?

 Answer

The Per base sequence quality report shows that trimming the last 10 bases may be a good idea.

Let's look at tools to do such manipulations to fastqc files, if we have to.