Help needed on Fastqc-adapter content

biotechkk · 07-29-2014, 09:22 AM

Dear all,

i have reads from nextseq 500. I have removed adaptors using fastq-mcf. I run Fastqc for the trimmed reads. I am getting fail result for adapter content,k-mer and per base seqeunce content.
Whether iam missing any adapters to remove?. I can see the probelm with first 30bp. I have attached adapter seq and screenshopt of fastqc results. Any help will be highly useful to me.

Thanks

Regards
Manoj

GenoMax · 07-29-2014, 09:35 AM

Is this RNAseq data? The bias seen at the beginning of reads is a common observation because of the "random" primers not being so random (http://seqanswers.com/forums/showthr...t=4846&page=14) and does not need trimming. A similar bias is also seen with tagmentation reactions and is normal.

biotechkk · 07-29-2014, 10:04 AM

Thanks. It is exon data. Can i remove first 20bp. Is there any other way to remove that bias. i have mapped with bwa and got mapping quality of 60%. So i need to remove this bias in the beginning of the reads.

GenoMax · 07-29-2014, 10:29 AM

Does the mapping quality (60%) refer to the percentage of reads that are mapping? I doubt the low % is because of the first 20 bp but if you want to give it a try you could remove them and see if it improves your mapping.

What kind of reference genome are you mapping to? Is it a reasonably complete one (e.g. human)?

More than likely you may have some other issue (primer dimers etc) that may be preventing reads from mapping.

biotechkk · 07-29-2014, 11:00 AM

Yes. only 60% of paired reads mapped to human genome(hg19). how to identify primer dimer in my reads and remove those from my reads.

simonandrews · 07-30-2014, 12:47 AM

There are two problems here. You have the typical RNA-Seq kmer bias problem at the start of your reads showing up in your Kmer and base composition plots. This isn't something you can fix, or that would be improved by trimming so don't worry about this.

Secondly you have read-through adapter contamination of your reads as shown by the adapter content plot however this is not produced by the normal common Illumina adapter, but comes from the transposase which Illumina uses to fragment their libraries in some of their kits. This means that your data would benefit from being trimmed to remove this and this will help with the mapping efficiency. You'll need to modify the default options for the trimming program you use to specify that it's the transposase sequence (CTGTCTCTTATA) which you want to remove. Pretty much all trimming programs will allow you to specify this sequence. After trimming you should see the adapter content plot be flat in the trimmed data and hopefully your mapping efficiency will go up too.

biotechkk · 07-30-2014, 02:11 AM

Thanks for your help. I have removed the transposase sequence and now i can see the adapter content plot flat. I will map the reads and see the maping quality.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
[FASTQC] Biases in GC whole sequence content	kazi1	Bioinformatics	9	06-03-2014 12:20 AM
FastQC,kmer content, per base sequence content: is this good enough	mgg	Bioinformatics	10	11-06-2013 11:45 PM
fastqc kmer content error	btaboada	Bioinformatics	1	08-13-2013 10:27 AM
FastQC: odd kmer content	zshuhua	Introductions	3	05-13-2013 08:36 PM
kmer content warning in FastQC	vallejov	RNA Sequencing	0	04-05-2013 11:10 AM

07-29-2014, 09:35 AM	#2
GenoMax Senior Member Location: East Coast USA Join Date: Feb 2008 Posts: 7,080	Is this RNAseq data? The bias seen at the beginning of reads is a common observation because of the "random" primers not being so random (http://seqanswers.com/forums/showthr...t=4846&page=14) and does not need trimming. A similar bias is also seen with tagmentation reactions and is normal.

07-29-2014, 10:04 AM	#3
biotechkk Junior Member Location: germany Join Date: Jul 2014 Posts: 5	Thanks. It is exon data. Can i remove first 20bp. Is there any other way to remove that bias. i have mapped with bwa and got mapping quality of 60%. So i need to remove this bias in the beginning of the reads.

07-29-2014, 10:29 AM	#4
GenoMax Senior Member Location: East Coast USA Join Date: Feb 2008 Posts: 7,080	Does the mapping quality (60%) refer to the percentage of reads that are mapping? I doubt the low % is because of the first 20 bp but if you want to give it a try you could remove them and see if it improves your mapping. What kind of reference genome are you mapping to? Is it a reasonably complete one (e.g. human)? More than likely you may have some other issue (primer dimers etc) that may be preventing reads from mapping.

07-29-2014, 11:00 AM	#5
biotechkk Junior Member Location: germany Join Date: Jul 2014 Posts: 5	Yes. only 60% of paired reads mapped to human genome(hg19). how to identify primer dimer in my reads and remove those from my reads.

07-30-2014, 12:47 AM	#6
simonandrews Simon Andrews Location: Babraham Inst, Cambridge, UK Join Date: May 2009 Posts: 871	There are two problems here. You have the typical RNA-Seq kmer bias problem at the start of your reads showing up in your Kmer and base composition plots. This isn't something you can fix, or that would be improved by trimming so don't worry about this. Secondly you have read-through adapter contamination of your reads as shown by the adapter content plot however this is not produced by the normal common Illumina adapter, but comes from the transposase which Illumina uses to fragment their libraries in some of their kits. This means that your data would benefit from being trimmed to remove this and this will help with the mapping efficiency. You'll need to modify the default options for the trimming program you use to specify that it's the transposase sequence (CTGTCTCTTATA) which you want to remove. Pretty much all trimming programs will allow you to specify this sequence. After trimming you should see the adapter content plot be flat in the trimmed data and hopefully your mapping efficiency will go up too.

07-30-2014, 02:11 AM	#7
biotechkk Junior Member Location: germany Join Date: Jul 2014 Posts: 5	Thanks for your help. I have removed the transposase sequence and now i can see the adapter content plot flat. I will map the reads and see the maping quality.