Solexa - same sequence but unique identifier

[Layla]

Hi all,

I am seeing this data from a paired end solexa run (pipeline 1.4). The sequence is identical, yet the identifiers (tile# on the flowcell, x and y coordinates) are different. In Bowtie alignment each sequence hits the same start and stop site.

First, is this PCR bias and second, should all the alignments be kept, none or only 1? Thank you for any comments.

@HWI-EAS261:6:2:675:1607#0/1
ATGTGCTCACCTGCCTCATCCATACACATACGTGGCTGCTCTCAC
+
ACC=BBAC@CBB@B?;=BAC;?B4;@7BBAA@?;@>A@;=9<3;@
--
@HWI-EAS261:6:7:1000:1185#0/1
ATGTGCTCACCTGCCTCATCCATACACATACGTGGCTGCTCTCAC
+
BCCBCCBCBBBCBCACABCABCBBABCBCCBB@B@BBB9B@A@=6
--
@HWI-EAS261:6:22:934:1507#0/1
ATGTGCTCACCTGCCTCATCCATACACATACGTGGCTGCTCTCAC
+
BC>ABBCBABCBCCCCC>BBABBCCBCBCCBB@BBB8B@A@@;<;
--
@HWI-EAS261:6:24:285:958#0/1
ATGTGCTCACCTGCCTCATCCATACACATACGTGGCTGCTCTCAC
+
BBB7BBA@B@@@?@A?AB@AAB@B@B@A?B@@=?8@=<<7;<<?7
--
@HWI-EAS261:6:60:421:1920#0/1
ATGTGCTCACCTGCCTCATCCATACACATACGTGGCTGCTCTCAC
+
BCB?;CBCCBBB;BA@BCBBBAABBBCCBB?B;;<9@:1)5<=@@
--
@HWI-EAS261:6:72:284:58#0/1
ATGTGCTCACCTGCCTCATCCATACACATACGTGGCTGCTCTCAC
+
<<A<ACCCCCACCBBCCCBAA<ACBBCBACBB>A<A6@>A@><76

Layla

[dawe]

I am seeing this data from a paired end solexa run (pipeline 1.4). The sequence is identical, yet the identifiers (tile# on the flowcell, x and y coordinates) are different.

Isn't this the rationale behind massively parallel sequencing? :-)
Many procedures rely on this: it allows greater confidence in (re)sequencing projects and allows "peaks" in ChIP-seq experiments...

[Layla]

Thank you for your quick response dawe
L

[kmcarr]

First, is this PCR bias...?

To answer that question you would need to calculate (estimate, make a rough guess at) the expected frequency of two independent reads starting at the same base and compare that to the observed frequency. Roughly speaking this relates the size of what is being sequenced to the number of reads collected. If you are sequencing a bacterial genome whose size is 2Mbp and you collect 20 million reads you would expect 10 reads starting at each base (assuming reads are perfectly distributed). If your target DNA size is larger than the number of reads collected than the expectation is that there should be few if any duplicate reads. The size of the "target" will of course depend on what your application is. Genomic sequencing, mRNA-Seq, Chip-Seq, etc.

I just re-read your post and saw that the data is from a paired-end run. What about the mates for the example reads, are they all the same sequence? If the sequences for two paired reads match for both read1 and read2 then it is almost certain that these reads are PCR duplicates.

[Layla]

Thank you for your reply kmcarr.

According to my calculations from medip-seq on the human genome, theoretically I should not have any duplicates

The mates of these reads have a different sequence

In the following scenario is it correct to maintain all the reads or should only 1 be kept from 0/1 (same id, start and seq) and all removed from 0/2 (ambiguity due to multiple start sites?)

HWI-EAS261:6:1:836:220#0/2 - chr4 151950446 GATANGATACGATACGATACGATACGATACGATACGATACGATAC
HWI-EAS261:6:1:836:220#0/2 - chr4 151950441 GATANGATACGATACGATACGATACGATACGATACGATACGATAC
HWI-EAS261:6:1:836:220#0/2 - chr4 151950476 GATANGATACGATACGATACGATACGATACGATACGATACGATAC
HWI-EAS261:6:1:836:220#0/2 - chr4 151950481 GATANGATACGATACGATACGATACGATACGATACGATACGATAC
HWI-EAS261:6:1:836:220#0/2 - chr4 151950486 GATANGATACGATACGATACGATACGATACGATACGATACGATAC
HWI-EAS261:6:1:836:220#0/2 - chr4 151950491 GATANGATACGATACGATACGATACGATACGATACGATACGATAC
HWI-EAS261:6:1:836:220#0/2 - chr4 151950456 GATANGATACGATACGATACGATACGATACGATACGATACGATAC
HWI-EAS261:6:1:836:220#0/2 - chr4 151950461 GATANGATACGATACGATACGATACGATACGATACGATACGATAC


HWI-EAS261:6:1:836:220#0/1 + chr4 151950305 AAGTGGCCAGGAATGGTTGTGTGTGCCTGTGGTCCCAGCTAGGTC
HWI-EAS261:6:1:836:220#0/1 + chr4 151950305 AAGTGGCCAGGAATGGTTGTGTGTGCCTGTGGTCCCAGCTAGGTC
HWI-EAS261:6:1:836:220#0/1 + chr4 151950305 AAGTGGCCAGGAATGGTTGTGTGTGCCTGTGGTCCCAGCTAGGTC
HWI-EAS261:6:1:836:220#0/1 + chr4 151950305 AAGTGGCCAGGAATGGTTGTGTGTGCCTGTGGTCCCAGCTAGGTC
HWI-EAS261:6:1:836:220#0/1 + chr4 151950305 AAGTGGCCAGGAATGGTTGTGTGTGCCTGTGGTCCCAGCTAGGTC
HWI-EAS261:6:1:836:220#0/1 + chr4 151950305 AAGTGGCCAGGAATGGTTGTGTGTGCCTGTGGTCCCAGCTAGGTC
HWI-EAS261:6:1:836:220#0/1 + chr4 151950305 AAGTGGCCAGGAATGGTTGTGTGTGCCTGTGGTCCCAGCTAGGTC

[Xi Wang]

it seems that your sequence 0/2 is mapping to a simple repeat region. I think you can keep one of them. but it largely due to what your application is.