Trimming low quality bases
There are a number of open source tools that can trim off 3' bases and produce a FASTQ file of the trimmed reads to use as input to the alignment program.
The FASTX-Toolkit provides a set of command line tools for manipulating fasta and fastq files. The available modules are described on their website. They include a fast fastx_trimmer utility for trimming fastq sequences (and quality score strings) before alignment.
FASTX-Toolkit is available via the TACC module system.
Let's run fastx_trimmer to trim all input sequences down to 90 bases:
- The -l 90 option says that base 90 should be the last base (i.e., trim down to 90 bases)
- the -Q 33 option specifies how base qualities on the 4th line of each fastq entry are encoded. The FASTX toolkit is an older program, written in the time when Illumina base qualities were encoded differently. These days Illumina base qualities follow the Sanger FASTQ standard (Phred score + 33 to make an ASCII character).
Exercise: fastx toolkit programs
What other fastx manipulation programs are part of the fastx toolkit?
Type fastx_ then tab to see their names
See all the programs like this:
Exercise: What if you just want to get rid of reads that are too low in quality?
Let's try it on our data- trim it to only include reads with atleast 80% of the read having a quality score of 30 or above.
Exercise: Compare the results of fastq_trimmer vs fastq_quality_filter
Unlike general fixed-length trimming (e.g. trimming 100 bp sequences to 40 or 50 bp), adapter trimming removes differing numbers of 3' bases depending on where the adapter sequence is found.
The GSAF website describes the flavaors of Illumina adapter and barcode sequence in more detail https://wikis.utexas.edu/display/GSAF/Illumina+-+all+flavors
One of the programs available as part of the fastx toolkit does a crude job of clipping adaptors out of sequences.
fastx_clipper will clip a certain nucl. sequence (eq: adapter) from your reads.
The cutadapt program is an excellent tool for removing adapter contamination. The program is not available through TACC's module system but we've installed a copy in our $BI/bin directory. Cutadapt has some advantages over fastx_clipper:
- Cutadapt allows for mismatches.
- Cutadapt allows for paired end support.
When you run cutadapt you give it the adapter sequence to trim, and this is different for R1 and R2 reads.
- The -m 22 option says to discard any sequence that is smaller than 22 bases after trimming. This avoids problems trying to map very short, highly ambiguous sequences.
Paired end commands are a little more complicated: It's a multi step process: first providing both R1 and R2 files to create tmp files which are input to the second cutadapt command.
**** NOTE: If the $BI/bin cutadapt doesnt work, use the one installed in my work directory. To use it, do this first:
Please refer to https://wikis.utexas.edu/display/GSAF/Illumina+-+all+flavors for Illumina library adapter layout.
Below information holds true from old ligation-based RNA-libraries. In this case, the R1 adaptor is different from the R2 adaptor. For current stranded RNA libraries (such as dUTP, used by GSAF), the R1 adaptor would be the same as the R2 adaptor. Fastqc reports will give you an idea about which adaptor is present in your data. Further, it's always a good idea to grep -c <partofdaptorseq> <fastqfile> to make sure you have the right adaptor sequence before trimming.
BACK TO COURSE OUTLINE