Good evening,
The multiqc report shows the presence of poly- tails in my samples after DNA (!) sequencing even after cleaning with fastp and cutadapt with appropriate settings. I think these are sequencing artifacts and I want to find reads with such "polyA-tails" in fastq.gz files.
Could you help me, please: how long should the sequence AAA... at the end ot read be so that fastqc and then multiqc consider it as a polyA-tail? This is necessary for search.
Thank you in advance, Poecile
P.S. By the way, these polyA-tails appear on adapter charts only after combining samples using multiqc, they aren't exist on fastqc charts for individual samples.
•
written
by
poecile.pal •
Don't trim polyA tails unless to problem seems to be acute
most aligners can align just fine, soft clipping the trailing As if those were present
long runs of As can naturally occur in transcriptomes and you run the risk of altering real data
At least 12 A's (AAAAAAAAAAAA)
Note that if the DNA fragment is shorter than the read length, as can happen with miRNA sequencing, Illumina will report the empty trailing space as a string of low quality A's
Another thought, is are these actual artifacts??? You can search for your Poly-A's with BBduk, as the previous poster mentioned 12 A's, you can make a 12 A's fasta
> 12 Poly A
AAAAAAAAAAAA
Use that fasta you created to separate reads with at least 12 A's. From that file, you can copy the sequence to NCBI Blast to find if it actually matches to a sequence in the archive.
I've done the same thing on my DNA reads. Novaseq 6000 is known for having Poly-G Poly-C tails, but then is everything with these tails artifacts??? Probably not.... Especially if I take the sequence and it maps at close to 100% match with a species, if they are paired reads, you can find the matching pair and see if it matches the same species. So when I find these that match a species, I will download the genome of that species and map that species and set them aside, before I actually start chopping off the ends.
I actually do other things a little different as well. My samples are human blood samples that shouldn't have any pathogens. So when and if I can find any pathogen that provides a good assembly, and the contigs match nearly perfect (ideally perfect), then before I do anything with my reads (adapters or QC trimming), I will first map my reads at something between 98-100%, and set these aside first. Then I will do the same thing after any other trimming or adapter removal, this will provide the longest reads.
Another option is to use Kraken2 with default settings and if I find any species with a significant number of hits, I will extract those and try to assemble them with Metaspades (with or without some host removal). Then if I can find a matching strain, I will download that Genome and map to that as well.
Log in to answer this question.
If this is DNA sequence then there should be no poly-A's in the sense of poly-A tails.
Use
bbduk.shin filter mode withliteral=AAAAAA(adjust the length as needed). If you need the poly-A's to be only in 3'-end of the reads then addrestrictright=NNparameter.Please show an example plot/table.