SAM file and unpaired reads

[Deb]

I've been using bowtie to map reads and RSEM for expression analysis. I posted this on the RSEM listserv but didn't receive any helpful responses. I thought I might have better luck here.

I have a sam file generated with bowtie using paired-end reads. When I try to run it through 'convert-sam-for-rsem' I receive the following error: "Number of first and second mates in read HWI-ST261:376:AD0D1MABXX:1:1:10000:106304:R1's full alignments (both mates are aligned) are not matched"!

From my understanding, this means that both mates in the pair are not present in the file. However, when I looked for the particular reads in my file, I find the following:

HWI-ST261:376:AD0D1MABXX:1:1:10000:106304:R2 163 comp3699_c0_seq2 1774 255 93M = 1837 156

HWI-ST261:376:AD0D1MABXX:1:1:10000:106304:R1 83 comp3699_c0_seq2 1837 255 93M = 1774 156

Both mates are present and the starting coordinate for each match. I want to write a script to remove the problematic reads but am not sure what 'convert-sam-for-rsem' considers an unmatched read.

Thanks for your help.
Deb

[Richard Finney]

Look at your sam flags (second field)
Binary for 1774 is 11011101110
Binary for 1837 is 11100101101
First bit is at the far right.
See explanation for bit flags at:
http://samtools.sourceforge.net/SAM1.pdf

Starting at the first bit we can see there's a problem. Bit1 (first bit) is contradictory.
"Read is paired" ... but ... "it's not paired" (???)

You can use this website : http://picard.sourceforge.net/explain-flags.html
to help you figure out what 1774 and 1837 mean.

The flags don't make any sense.
first read is ...
Summary:
read mapped in proper pair
read unmapped
mate unmapped
mate reverse strand
first in pair
second in pair
read fails platform/vendor quality checks
read is PCR or optical duplicate

Second read is ...
Summary:
read paired
read unmapped
mate unmapped
mate reverse strand
not primary alignment
read fails platform/vendor quality checks
read is PCR or optical duplicate

You may want to doctor the flags or filter reads with funny flags.
Playing around you can figure out what's best.
Not all sam output producing software takes these flags seriously; but some software that takes sam as input does take it seriously. Making them play nice takes a bit of dirty work.

I'm not sure if there's an additional problem of ... the software can't figure out that ****:R1 read pairs with ***:R2.

[Deb]

Thanks for your reply Richard but I think you are confusing fields. My bit flags are 163/83. The fields that you listed (1774/1837) should be the positional coordinates, unless I am misreading it.

Thanks for the explain-flags site; I wasn't familiar with it.

[kmcarr]

Deb,

The problem is the :R1 and :R2 extensions to the names. The SAM spec states the the names of both mates should be identical. Read 1 or read 2 status is recorded in the flag.

[Deb]

Thank you to you both! I am going to try remapping without the :R1 and:R1 to see if that solves my problem.

[Richard Finney]

yep. I missed the flags field completely. 163 and 83 do make sense.
Cutting off R1 and R2 will probably do the trick for you .