Question: Weird repetitive sequences in metagenomic WGS reads and assembled contigs
gravatar for shlee354
16 months ago by
shlee3540 wrote:

Hi all,

I'm working with some metagenomic WGS data of induced sputum. Here's the sequencing information of the samples:

  • Type of sequencer: Illumina platform
  • Type of reads: Paired-end
  • # of reads: ~200M per sample
  • Read length: 151
  • Library kit: TruSeq Nano DNA kit

I used Trimmomatic to remove adaptor sequences and low-quality ends and then used KneadData to remove the human contaminant. Since the airway is almost sterile, ~99% of reads were discarded during the process. Remaining decontaminated reads were then assembled into contigs(scaffolds) using metaSPAdes, and microbial genes were called from them using Prodigal.

The problem is:

While inspecting assembled scaffolds, I noticed that there are contigs with a bunch of weird repeating sequences. Like:


or like:


To see the proportion of these strange contigs, I calculated the fold change of the observed frequency of every possible tetranucleotide to the expected value of it (1/256 in this case) for all contig.

FC of a tetranucleotide frequency to the expected value = 
(# of the tetranucleotide / # of the total tetranucleotide in the sequence) * 256

The following is the histogram of top FC of each contig. It shows clear bimodality: Tetranucleotide FC distribution of my data (Peak on the left is located about 3.5, and the other, about 28~29.)

And this is the tetranucleotide FC distribution of contigs made from gut microbiome WGS data. Tetranucleotide FC distribution of normal data

I did the same on the decontaminated reads, but trinucleotide instead of tetranucleotide (since the reads are only 151bp long). It also showed clear bimodality: Trinucleotide FC distribution of my decontaminated reads

And reads before running KneadData seems normal, compared to the gut metagenomic reads: (Top: my trimmed reads, bottom: gut metagenomic reads) Trinucleotide FC distribution of my trimmed reads Trinucleotide FC distribution of gut metagenomic reads

During the manual inspection of FASTQ files, I also found strange reads that seemed to be assembled to abnormal contigs:


The question is:

Where did these strange readings come from? Is this an error in the Illumina platform itself? The quality score itself does not seem to be bad that much. Any helps would be appreciated.


wgs metagenome assembly • 318 views
ADD COMMENTlink written 16 months ago by shlee3540

Repeats are very common in genomes, and therefore it is not surprising to see them. My guess is that they are in part from human DNA. You could try RepeatMasker or Dfam even though there is nothing coming up for your examples.

ADD REPLYlink modified 16 months ago • written 16 months ago by Michael Dondrup48k

I'll try it. Thanks, Mr. Dondrup.

ADD REPLYlink written 16 months ago by shlee3540
Please log in to add an answer.


Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 1922 users visited in the last hour