Illumina Mate Pair prep

[Protaeus]

Hello,

I was writing to ask a question of those who have, seemingly successfully, worked with Illumina mate pair data. I have seen other threads in which people mention that reverse complementing the reads is a necessary prerequisite to ensure the reads are facing the correct direction. Is this true? If so, is there a tool that someone could recommend to easily reverse complement a fastq or sequence.txt file?

If you have any other suggestions for dealing with mate pair data and have suggestions of tools to accomplish those tasks, I'd greatly appreciate it.

Cheers,
John

[Jon_Keats]

Use the fastx-toolkit. It has a great reverse-complement script that works very well. And yes if you want to analyze in a pseudo-paired-end analysis with aligners like BWA this is a needed step. Seems to work pretty well, also tried using Novoalign which doesn't require the reverse-complement step but is very slow in comparison and only marginally improves the outcome.

[bronnyd]

I'd like to be sure I understand the orientation of paired-end reads for Illumina.

Imagine you ignored the middle and wanted to concatenate the paired-ends together so that they read 5'-3' from the same strand. You leave PE1 as it is. Do you concatenate the reverse or the reverse compliment of PE2?

The help would be great. Thanks all.

[kmcarr]

I was writing to ask a question of those who have, seemingly successfully, worked with Illumina mate pair data. I have seen other threads in which people mention that reverse complementing the reads is a necessary prerequisite to ensure the reads are facing the correct direction...

Whether or not you need to rev-comp one of the mate reads would depend on what alignment program you are using. I can tell you that Bowtie can deal with the sequences directly, you just have to tell it what to expect. There are three options you can use with bowtie:

--fr  Appropriate for Illumina paired end reads (default)
--rf  Appropriate for Illumina mate paired reads
--ff  Appropriate for SOLiD paired reads (becomes default when you specify colorspace reads using the -C option)
.

If you plan to use bowtie to align your mate paired reads you do not have to rev-comp anything, just use the --rf option when you run bowtie.

[Protaeus]

any clue if I would need to do so with bfast?

[bronnyd]

Great, the assembly programs know how to handle Illumina PE reads. But how should I understand them when I see two paired end reads on a screen:

>PE1
CTTACCCC
>PE2
ACCTAAAA

If one strand of one insert were like a sentence, do I read it from left to right on PE1, skip over the unknown stuff, and finish the insert from right to left on PE2 (like this)?

CTTACCCC-----------------------AAAATCCA

or is it actually the rev-com of PE2 (like this)?

CTTACCCC-----------------------TTTTAGGT

Thanks!

Protaeus - I apologize if my question belongs in another post! Thanks for bringing up PE orientation today.

[Protaeus]

Hey Bronnyd,

No problem, though hopefully the responders make it clear to whom they're responding! The mate pair prep and the paired end prep are quite different. In general, I don't think you have to worry about orientation with paired end data, at least with most alignment tools. The mate pair differs significantly in prep and results in some orientation issues that I think must be pre-processed prior to alignments. Guess we'll find out...

John