Thank you very much for your answer. The command I used is: cutadapt -a SEQ --trim-n -o out.fq in.fq, I used a perl script to remove reads < 18-nt. Can I ask a further question? Although the pairing information is lost as you mentioned, are the mapping positions of reads correct? To me, the mapped positions of reads are important in following analysis. Thanks!
Removing shorter reads in PE mapping
Dear all,
I have pair-end (PE) sequencing data (R1.fq and R2.fq). First, I removed the adapters, which resulted in unequal sequence length for some PE reads. Then I removed the reads with length < 18-nt, which resulted in different number of reads for the R1.fq and R2.fq data. Do you think it OK to continue tophat2 mapping? (I am trying this, and it seems OK)
I appreciate any of your comments. Thanks!
1 answer
1) please do tell us how (the commands) you used to trim adapters and remove short reads.
2) No, if R1 and R2 have different number of reads, probably the pairing information is lost and your mapping will be wrong, as tophat will assume a non-existing pairing between R1 and R2.
are the mapping positions of reads correct
Most probably not. If you are taking out any reads from one file, its mate should be taken out of the corresponding (R1/R2) file to keep the order of reads consistent in files. Aligners do not check to ensure that the read pairs are in order.
Log in to answer this question.
More posts like this
-
Trimming Nextera adapter from scRNA paired reads with different length
written by almogangel •
-
what can i do with R2 reads
written by yunlongli029 •
-
How can I extract the overlap of the sequence data
written by 190444373 •
-
How to process the raw reads from miRNA-seq
written by smallfish •
-
Any tool to output unmapped or mapped_with_clips reads from a BAM to two fastq files for PE data?
written by chen
-
pair-end, mate-pairs coding
written by kamiljaron
-
paired end illumina reads
written by midox
-
Tophat --max-multihits option
written by biolab
-
tophat --library-type option
written by biolab
-
One question about using bowtie2 in genome mapping
written by zhhxu9 •
For normal fragment libraries, adapter-trimming read pairs should leave them the same length. It's important to process paired reads together with a pair-aware program to prevent these issues of differential length after adapter-trimming and mismatched reads.
Thank you very much, Brian. So, do you think one way is to do separate mapping for R1 and R2? Thanks!
Yes, that is one way to do it, but the reason to run paired reads is because the mapping is more accurate, so you'll end up with inferior results. Also, you'll end up with untrimmed adapters by not doing adapter-trimming together. I recommend deleting your trimmed files and following this approach, starting from the raw fastqs. Specifically:
...where adapters.fa is in bbmap/ref/
Thanks a lot Brian. It's very helpful. I will try that.