SAMTools finds PCR duplicates based on start and end position (in case of paired-end reads). I have a dataset of single end reads of variable length. Will SAMTools rmdup work on these to find out PCR duplicates? If not, then will seeding each nucleotide one by one able to find the duplicates? If yes then is there any tool available based on seeding principle?
what makes you shure that your duplicates with different starts and the same end positions (or vice versa) are actually real pcr duplicates? I cannot imagine a scenario how this actually may happen in your pcr tube... Or did you per chance use local alignment with trimming? In that case I would do end-to-end alignment, which would probably discard the reads that were trimmed before, and then you can to rmdup without problems. What was your sequencing template and what are you looking for?
Hi Stephan. I think there is some misunderstanding about the topic. While dealing with paired-end read, PCR duplicates can be pointed out by looking at the start and end position. As per my understanding samtools obey this logic while executing rmdup. But, within single-end read dataset, finding out PCR duplicates is bit tricky. So, my question is is there any available software/tool which can find PCR duplicates, while seeding each nucleotide into reference genome.
Looking for your suggestions.
rmdup samtools rmdup [-sS]
Remove potential PCR duplicates: if multiple read pairs have identical external coordinates, only retain the pair with highest mapping quality. In the paired-end mode, this command ONLY works with FR orientation and requires ISIZE is correctly set. It does not work for unpaired reads (e.g. two ends mapped to different chromosomes or orphan reads).
OPTIONS:
-s Remove duplicate for single-end reads. By default, the command works for paired-end reads only.
-S Treat paired-end reads and single-end reads.
Did you try -s?
May this discussion is helpful
http://seqanswers.com/forums/showthread.php?t=5424
Sorry for replying late. Yes I have tried '-s' option while executing samtools rmdup. Almost 20% of the reads were taken out as pcr duplicates. Also I am not getting any idea by searching literature that how it is working.
And also thanks for the seqanswers link. It helped me to understand the issue in details.
PCR duplicates certainly don't have exact sequence and also necessarily all should not end at same postion.
So let me modify my question: Is there any tool where I can specify the seed length to consider any read as pcr duplicate. I am having single end reads of variable lengths.
And now definitly I will try with picard for sure.
Thanks in advance.
"PCR duplicates certainly don't have exact sequence and also necessarily all should not end at same postion."
Why should PCR duplicates not have the same position when you use end-to-end alignment? Even with local alignment it should yield the same position in most cases...
If reads map to the same locus with slightly shifted start/end positions I would assume that these are independent reads, originating from different template copies, and mapping by chance to the same locus. How often this happens depends of course on your sequencing depth... Could you explain why you think that reads with slightly shifted mapping coordinates are PCR duplicates?
I am not talking about shifting of mapping coordiantes. Suppose I have two reads (remember I am dealing with single end reads) of length 78 and 100 respectively and suppose 15 bases of second read has been trimmed off due to low quality. Now length two reads has truned into 78 and 85 respectively. So while mapping , say 5'-end of both reads mapped to identical position and 7 nts is in overhang conditon w.r.t. first read but mapped. Can't these two reads be considered as PCR duplicates.
Thanks in advance.
Ok, now I understood what you mean. I am not shure that these are PCR duplicates, since they should be fairly similar in quality and length. In my data the read lengths are almost always 99-101 bp in a 100 bp run... So I am curious why you would get so different read lengths on a 100 bp run. What is your sequencing template?
Apart from that, I do not think that there is a ready-to-use software package available to solve your problem. If you had a program that uses seeds to determine duplicates, how would that program differentiate between duplicates of different lengths and independent overlapping reads - how would you descriminate both?
Maybe you could write a script that does that kind of duplicate removal.
First you would need to get reads with the same starting coordinates from your sorted BAM file - that should be easy. But what would you do with those 2 reads? In your example: would you trim the longer read (and sacrifice coverage) and create a consensus read - or would you discard the shorter read? What if there are sequence differences between both - e.g. substitution in 1 read - would you create an ambiguous consensus?
Hi Stephen. Ion torrent single end read dataset is usually consists of reads of variable length. The dataset I am handling is having reads from 5 nts to 370 nts long. That is why I need an algorithm to finding out PCR duplicates or uniquely mapped reads.
I really don't have any answer for rest of your question. We still have to figure it out.
Still awaiting for some more ideas.
Thanks in advance.
you may find this post very interesting:
http://seqanswers.com/forums/archive/index.php/t-6854.html
There is a discussion going on wether it is at all necessary and always wise to do PCR duplicate removal - it all depends on the question of your experiment and your library template... What material are you actually sequencing?
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Bioinformatics
- Bioinformatic Tools