Unexplained Unmapped Read Mates Produced By Tophat2

Question

[mathieu.bahin]

Hi all,

I have a SAM file produced by TopHat 2 (version 2.0.8) in which I find these two following mates that I can't explain:

HISEQ4_0101:8:1101:10030:81268#TGACCA   73      17      22696364        50      67M     *       0       0       ATGGATATGAATTTTTTGCTAAACGACAATTGGTAACCCTATTTTCAGCCCCAAATTACTGTGGCGA  CC@FFFFFHHHHHJJIJJJJJJJJIDGEIEIJDBDGICHCGIJJIJIJIJJIHJJIGHIEIHEFCD?     MD:Z:67 RG:Z:1  XG:i:0  NH:i:1  NM:i:0  XM:i:0  XN:i:0  XO:i:0       AS:i:0  XS:A:+  YT:Z:UU

HISEQ4_0101:8:1101:10030:81268#TGACCA   153     17      22696364        50      101M    *       0       0       ATGGATATGAATTTTTTGCTAAACGACAATTGGTAACCCTATTTTCAGCCCCAAATTACTGTGGCGAGTTTGATAATGCTGGTGGAATGATGAGTGTGGAT        DDEDDEEDDDBDBDDDDDDDDDDDEEEFFEFFFHE?FIHGGGHGJJIJJIIHIJGHDBJJJIGHHHCIIIIIJIJIIIJJJIJJJJJIHHGHGFFFFDBB<        MD:Z:101        RG:Z:1  XG:i:0  NH:i:1  NM:i:0  XM:i:0  XN:i:0  XO:i:0  AS:i:0  XS:A:+  YT:Z:UU

• Same read ID
• the flags show a first (73) and a second (153) in pair
• Mapped on the same chromosome
• Mate unmapped (7th field = "*")
• NH:i:1 for both (no multiple alignment)

Why is there a star in the 7th field instead of an equal as expected ?

Cheers

[Devon Ryan]

You might check what settings tophat2 is using when it's running bowtie2. That might be considered a "dovetail" mapping in bowtie2, so it wouldn't be valid for mates by default (try feeding that to bowtie2 by itself while playing with the settings to check this).

Answer 1

[Pierre Lindenbaum]

there is a big problem with tophat2:

• tophat : sam-flag 115 = properly-paired + read.reverse + mate.reverse ? + http://seqanswers.com/forums/showthread.php?t=36055 + https://groups.google.com/d/topic/tuxedo-tools-users/jstOyQ-zRgs/discussion
• " Tophat reports discordant sam entries for concordant pairs " https://groups.google.com/d/topic/tuxedo-tools-users/ASWQqS2xje4/discussion

And I'm still waiting for an answer...

[Devon Ryan]

Did you ever try feeding a few of those reads (only non-spliced ones) into bowtie2 while playing with the --dovetail option? I suspect that it's bowtie2 that's actually the issue here.

[Pierre Lindenbaum]

I tried to align the two reads with bowtie2 only (+default opts). The SAM flags were OK.