Directionality in targeted resequencing

[Harsh Goswami]

Hi all,

I am just a kid in NGS. Would anybody be kind enough to explain me what is "directionality" in a targeted resequencing experiment? Why is it important?

[CPCantalapiedra]

Well, I am a kid in NGS too, hehe but as no adult is answering...

I am not very sure that directionality is such important for targeted resequencing, altough it seems clear that would facilitate unambiguous mapping. In the case of RNAseq, that could be seen as a special case of targeted sequencing, this is from the Illumina webpage "Maintaining transcript strandedness can enhance the value of RNA-Seq for many applications such as transcriptome annotation and bacterial transcriptome profiling. Generate directional libraries for large numbers of high quality reads using the directional (strand-specific) mRNA-Seq Sample Prep protocol." This is important since mRNA is single stranded but cDNA is not. Thus, if this were the same for targeted resequencing, directionality would be synonym of strand knowledge; altough often terminology is unclear. Please, consider this answer more as a way to prompt people with experience than to give an accurate answer hehe

[Harsh Goswami]

Many thanks CPCantalapiedra for some clue.

[Harsh Goswami]

Please through some more light on this topic...Please !!

[arvid]

Depending on the application, it is often interesting in RNA-Seq to be able to distinguish if a read comes from a sense (i.e. protein coding) or antisense (possibly regulatory) transcript. Directional protocols, which preserves the strand identity of the original RNA molecule as CPCantalapiedra pointed out, usually result in lower yields and could cause other problems (biases) as well, however this is outside my expertise.
However, targeted resequencing is typically done on DNA, so for that I'm not quite sure why directionality would be important...

[Heisman]

Perhaps you are referring to the fact that at the ends of targeted intervals you typically only get coverage on one of the strands? This is a known strand bias with targeted resequencing.

[CPCantalapiedra]

Heisman, could you provide ref for that please? I would like reading more about this topic

[Heisman]

You can see page 7 here: ftp://ftp.sanger.ac.uk/pub4/resource...anual-1.01.pdf

As for why, I've busted out the microsoft paint skills to try to illustrate this. If the horizontal black line is genomic DNA, the green line is a targeted region, and the brown line is a bait, if the vertical black lines denoted a sonicated fragment you can see that you can pull down that fragment and sonicate the region in the red box from one end. If the vertical blue lines denoted a sonicated fragment, you would not pull that region down, and so you wouldn't sonicate the region in the red box from the other end. Hence, you have strand bias at the boundaries.

[CPCantalapiedra]

mmm not sure to understand. When you say "pull down" you mean capturing the fragment as single stranded with the bait? Neither I understand what you mean with sonicating the region from one end.
Anyway, thank you for the answer, and the drawing

[Heisman]

That is what I mean by pull down.

Sorry, I meant "sequence" the region, not "sonicate".

[CPCantalapiedra]

Does this means that one strand can be sonicated in a different position than the other strand? I thought they would be splited from the same point/nucleotide

[Heisman]

No, sorry, they will sonicate in the same place, at least roughly.

I'm not sure how to explain this beyond the picture, so I'll try again and type it out slightly differently. Let's say there are two copies of that genomic DNA template and one gets sonicated at the vertical blue lines and one gets sonicated at the vertical black lines. You're interested in the sequence in the red box. With the blue fragment, the red box will only be sequenced from the right. With the black fragment, the red box will only be sequenced from the left. As you are targeting the green segment via the placement of the brown bait, you will never pull down the blue fragment. You will pull down the black fragment. So, you will only sequence it from the left, and whichever strand that corresponds too.

[CPCantalapiedra]

thank you, I understand now! I would need to capture the blue fragment to get the other strand sequenced. Is this a general bias? I mean, longer paired-end reads improves this?

[Heisman]

Longer reads would improve it.

[CPCantalapiedra]

ty very much