Losing most PE reads using Cutadapt on 150bp PE metagenomes

Question

[lintonf]

Hi everyone,

I am using cutadapt to cut some adapters on my PE, metagenome reads (150bpx2, Illumina NovaSeq).

Here is my code:

for i in *_R1_001.fastq;
do
    SAMPLE=$(echo ${i} | sed "s/_R1_\001\.fastq//")
    echo ${SAMPLE}_R1_001.fastq ${SAMPLE}_R2_001.fastq/n
    cutadapt --pair-adapters -pair-filter=both --discard-untrimmed -b ACTGCGAA -B TTCGCAGT -o ${SAMPLE}_R1_001_cut.fastq -p ${SAMPLE}_R2_001_cut.fastq  ${SAMPLE}_R1_001.fastq ${SAMPLE}_R2_001.fastq**

done

I've done this a few ways (keeping untrimmed sequences, not keeping them, etc) - and for whatever reason, I am keeping only 0.2% of my reads (will attach picture of results).

I ran the sequences through FastQC and every single metagenome is great quality - so I am not sure why I am losing so many reads!

Does anyone know why this could be happening, and how to improve the reads I am keeping?

I was thinking of trying trimmomatic but I am not sure how to go about using custom adapters for trimmomatic.

[Istvan Albert]

you are discarding untrimmed reads, perhaps your reads don't contain the adapters

Answer 1

[h.mon]

As Istvan Albert said, you are discarding untrimmed reads, so all reads that do not contain the adapter will be removed - in general, this is useful when one is removing primers, not adapters.

Besides, it seems to me the sequences -b ACTGCGAA -B TTCGCAGT are barcodes, not adapters. Unless these are inline barcodes, they were sequenced during specific cycles in Illumina machines, are used to demultiplex a sequencing run, and are output to separate fastq files.

[lintonf]

Thank you for pointing that out, you're right.

I am trying to use bbduk to remove the adapters w/ their respective indexes (before I was clearly just looking for indexes) - hopefully this works.

Do you have another recommendation for removing the adapters with their indexes? FastQC says there's a universal Illumina adapter attached on the 3' end but otherwise the sequences are good quality. Some samples show that they have some overrepresented sequences (the adapters + indexes) and others don't, so I am not exactly sure how to correctly proceed.

Thanks for your help and guidance!

EDIT: used bbduk.sh to get rid of adapters with imbedded indexes and it was successful! Double checked my output with FastQC. Thanks again for your input!

[h.mon]

BBDuk is fine and should eliminate all adapters, if this is not happening, then you can post your command so we can check what you are doing.