[Bio-bwa-help] bwasw and picard fixmate problems.

[Louis L. <lou...@ma...>]

I'm aligning paired reads using bwasw but I'm getting results that are
surprising to me.

For example
read1 aligns well
read2 is split in 2. read2_1 has ~130bp that aligns somewhere, and
read2_2 has 45bp that align somewhere else.

read1 has RNEXT and RPOS set to '*'
1ead2_1 has RNEXT and RPOS set to the read1 values
1ead2_2 has RNEXT and RPOS set to the read1 values

So picard doesn't like this at all since read1 doesn't have the mate ref
set. It gives 'Mapped mate should have mate
reference name'

but shouldn't read 1 have RNEXT to read2_1 and read2_1 have RNEXT to
read2_2?

In anycase if I run fixmate on this (after querysorting everything) it
get's worst.

read1 and read2_1 now have their ref set to '*' with a position. BUT the
flag unmapped is not set. This creates an invalid BAM/SAM file.

If bwa creates these weird entries and fixmate can't fix them, how
should I align paired reads?

Here are the pre fixmate entries (I removed the actual sequences):
In this example I tried with the bwasw -M option to flag the 2nd read as
secondary to remove it afterwards
readName	129	chr1	64636	13	158M	*	0	0	XE:i:4	XF:i:3	RG:Z:NA_71_1
NM:i:0	AS:i:158	XS:i:150
readName	81	chr1	66404	28	39M7I79M5I51M2D44M25S	=	64636	-1983
XE:i:6	XF:i:3	RG:Z:NA_71_1	NM:i:15	AS:i:166	XS:i:148
readName	337	chr19	108166	6	205S45M	chr1	64636	0	XE:i:1	XF:i:1
RG:Z:NA_71_1	NM:i:0	AS:i:45	XS:i:34

Here is the post fixmate after I removed the secondary alignments
readName	81	*	66404	28	39M7I79M5I51M2D44M25S	*	64636	-1983	XE:i:6
XF:i:3	RG:Z:NA_71_1	NM:i:15	MQ:i:13	AS:i:166	XS:i:148
readName	161	*	64636	13	158M	*	66404	1983	XE:i:4	XF:i:3
RG:Z:NA_71_1	NM:i:0	MQ:i:28	AS:i:158	XS:i:150


Help and thanks
Louis