tophat paired reads mapped to different chromosomes

[wenhuang]

Hi,

I noticed that tophat some times can map paired end reads to different chromosomes (see one example below). The sequences are undoubtedly of low complexity.

I haven't looked to see how often this may happen (I guess rarely), but I wanted to ask if there is an option to control this behavior in tophat?

Thanks!


HWUSI-EAS787_0001:5:31:1709:802#TTTTTT 113 chr1 23 255 81M 62678829 0 AGATCAACTAAACAGAAAATTAACAAGGAAAAAAAAACAAAACAAAACATGTATCTATAAAGCTCACTAAAGCAAAGCACA GGHEHHHE@HHHGHHHHHHHHHHHHHECGGGGGGGGHGGGGHHHHHHHHHHHHHHHHHHHHHHFHHHHHHHHHHHDFFGGG NM:i:2
HWUSI-EAS787_0001:5:31:1709:802#TTTTTT 177 chr12 62678829 255 81M = 23 0 CTTTATAGATACATTTTTTGTTTTGTTTTTTTTTCCTTGTTAATTTTCTGTTTAGTTGATCTATCCATAAGTGTGAGTGGG #ECG@D9./9-@'7+:BCH,CCCB.CEEHHHHHDHDCG:GCE:HHHHFH5HGEB?GAAH9DHHHHHHHFHHHHHHHGHFHG NM:i:1

[choy]

Hi there -- I have noticed this behavior as well. To handle it, I filtered out the reads from the SAM file using SAMtools and pysam. I have notified Cole Trapnell about this behavior -- perhaps it will be corrected in a future version.

Cheers,
Matthew

[wenhuang]

I found out that there were not many cases like this.

I ended up filtering based on the bitwise flag in the second column of the SAM file simply using awk. Paired end reads mapping to different chromosomes falls in one of the flags that have improper mapped pairs. In addition to mapping to different chromosomes, there are cases where paired end reads are on the same strand, etc. This filtered out less than 10% of the aligned reads. I feel that these reads might be useful in looking for structural changes, but being able to control it is definitely good.

[snp_analyser]

still having the problem...

I filtered out alignments got from TOPHAT using bit 1 and 2 of the SAM flag (column 2), but am still encountering pairs mapped to different chromosomes. Any other way around this problem without actually checking for column 3?

Sample Output:

7_IR4IbRc7842 147 chr1 6917 255 75M = 7043 0 ACCTGCAAGATTAGGCAGGGACATGTGAGAGGTGACAGGGACCTGCAGGGGCAGCCAACAAGACCTTGTGTGCAC PQQQTNSOPPPTNTUQJR[QLPPPXNQKQNTTSTPTPZ^ZS\YTSTU^baa``baaaaaaab^abbbbbbbbaaa NM:i:0


7_IR4IbRc7842 99 chr9 7043 255 75M = 6917 0 CGGGGCCCAGGTCGGCAATGTACATGAGGTCGTTGGCAATGCCGGGCAGGTCAGGCAGGTAGGATGGAAAATGGA aaaaaaaaaaa\aaaaaaaa^aaa`a```Z^_RZ``^]^]]\][Y`[\^[JXV]XSPR]KP\BBBBBBBBBBBBB NM:i:2


By the way the sam flags translate to:
99=1100011
147=10010011