SEQanswers

SEQanswers (http://seqanswers.com/forums/index.php)
-   Epigenetics (http://seqanswers.com/forums/forumdisplay.php?f=24)
-   -   Directional vs. non-directional sequencing libraries (http://seqanswers.com/forums/showthread.php?t=34581)

daniel_g 10-17-2013 11:13 AM

Directional vs. non-directional sequencing libraries
 
Hello everyone,

My supervisor and I are working with bisulfite-treated whole genome sequencing data but we're having issues understanding the nuances of directional vs. non-directional libraries.

Specifically, in the Lister/Ecker paper published in 2009 detailing the MethylC-seq pipeline for analysing bisulfite-treated data (Finding the fifth base: Genome-wide sequencing of cytosine methylation), did they use a directional or non-directional library? Nothing in the paper seems to indicate which one was used.

Thanks,
Daniel

dpryan 10-17-2013 11:43 AM

The Lister/Ecker 2009 dataset is directional. When in doubt, just align the first 100,000 or so reads as non-directional and then look at the alignment statistics. If it's not actually non-directional, then the complementary strands will have very few reads aligned to them.

pmiguel 10-18-2013 07:29 AM

Quote:

Originally Posted by dpryan (Post 119101)
The Lister/Ecker 2009 dataset is directional. When in doubt, just align the first 100,000 or so reads as non-directional and then look at the alignment statistics. If it's not actually non-directional, then the complementary strands will have very few reads aligned to them.

"Complementary strands"? Do you mean the following?:

For a bisulfite-converted library, you would look some of the reads and see if all the conversions are C->T or G->A. If so, it is a stranded library. If the reads are mixed (about 1/2 have all C->T and the other 1/2 have all G->A conversions), then the library is not stranded.

--
Phillip

daniel_g 10-18-2013 07:32 AM

Okay great, thanks for the response! And yes, I see what you mean. I'm not actually working with their data directly but if I were that would be a quick way to tell.

pmiguel 10-18-2013 07:50 AM

The problem is if you are using some other method to look for 5-methylation of cytosines, it may not be possible. For example, if you use Me-DIP, then there is no conversion, you just see the unconverted sequence and assay for peaks.

It also kind of hurts my head to think that you can take a TruSeq DNA library prior to amplification, perform a methyl-cytosine immunoprecipitation on it. (This presumes you used non-methylated Y-adapters!) Then amplify and sequence it. This library will be directional! Because of the Y-adapters.

Not sure it gains you anything, but there you are...
--
Phillip

dpryan 10-18-2013 07:51 AM

Quote:

Originally Posted by pmiguel (Post 119220)
"Complementary strands"? Do you mean the following?:

For a bisulfite-converted library, you would look some of the reads and see if all the conversions are C->T or G->A. If so, it is a stranded library. If the reads are mixed (about 1/2 have all C->T and the other 1/2 have all G->A conversions), then the library is not stranded.

Sort of, though I can read that in ways that wouldn't be correct, so I'll clarify. If we C->T convert the reads (just read1 for paired-end data) and then align it to a bisulfite converted genome, the reads will be from a directional library if they only map forward on a C->T converted reference and reverse a G->A converted reference. If they also align appreciably in the reverse orientation on the C->T reference or forward on the G->A converted reference, then they very likely arose from a non-directional library prep. I could see someone misinterpreting what you wrote in the case of paired-end reads.

Since bisulfite treatment results in a loss of proper base-pairing between DNA strands, the new complementary oligos that are created in some library preps are then sometimes referred to as being complementary. Yeah, I can see where the confusion arises there!

dpryan 10-18-2013 07:53 AM

Quote:

Originally Posted by pmiguel (Post 119234)
The problem is if you are using some other method to look for 5-methylation of cytosines, it may not be possible. For example, if you use Me-DIP, then there is no conversion, you just see the unconverted sequence and assay for peaks.

Yeah, the whole directional/non-directional thing is specific to bisulfite treatment. When we start using nanopores that can directly read a base as methylated then this stuff will quickly become a historical footnote.

pmiguel 10-18-2013 07:54 AM

Quote:

Originally Posted by dpryan (Post 119235)
Sort of, though I can read that in ways that wouldn't be correct, so I'll clarify. If we C->T convert the reads (just read1 for paired-end data) and then align it to a bisulfite converted genome, the reads will be from a directional library if they only map forward on a C->T converted reference and reverse a G->A converted reference. If they also align appreciably in the reverse orientation on the C->T reference or forward on the G->A converted reference, then they very likely arose from a non-directional library prep. I could see someone misinterpreting what you wrote in the case of paired-end reads.

Since bisulfite treatment results in a loss of proper base-pairing between DNA strands, the new complementary oligos that are created in some library preps are then sometimes referred to as being complementary. Yeah, I can see where the confusion arises there!

Ah, I see. Thanks for the clarification.

--
Phillip

pmiguel 10-18-2013 08:04 AM

Quote:

Originally Posted by dpryan (Post 119237)
Yeah, the whole directional/non-directional thing is specific to bisulfite treatment. When we start using nanopores that can directly read a base as methylated then this stuff will quickly become a historical footnote.

And yet, the MeDIP libraries I describe above are, in fact, directional. But I don't think there is a way to detect that informatically after the fact.

But yes, we are talking the about technology, rather than science. (Neal Stephenson has a hilarious section of his book, Anathem, that highlights the difference.)

BTW, my guess is direct detection of methylated bases will be commercially available from Pac Bio before they are from Oxford Nanopore.

--
Phillip


All times are GMT -8. The time now is 10:44 PM.

Powered by vBulletin® Version 3.8.9
Copyright ©2000 - 2019, vBulletin Solutions, Inc.