Insulator is too biologically specific

jakebeal commented 6 years ago

From @jakebeal on August 2, 2016 13:31

Insulator is defined as SO:0000627, which is more biologically specific than the term is being used in practice.

Proposed in discussions: generalize this symbol, change its name to 'insulator/spacer' and submit a new SO term

Copied from original issue: SynBioDex/SBOL-visual-deprecated#12

shyambhakta commented 5 years ago

I agree. We ought to revive this.

SO:0000627 seems to only consider insulation from unwanted/aberrant transcription from neighboring "CRMs" = cis regulatory modules, and focuses on eukaryotic chromatin. For bacterial gene expression, that's just a terminator. The example we give of RiboJ doesn't satisfy the SO definition, because ribozymes don't insulate a gene from an upstream "CRM", like a promoter.

Modern bacterial gene expression insulators exceed that definition and directly remove context or initiate transcription or translation in such a way as to be context-nonspecific. Ribozyme insulators and RNase sites can provide insulation from 5′ and 3′ UTR context. Bicistronic design RBSs can provide insulation from CDS context. Spacers can, at least, provide insulation from DNA supercoiling from neighboring genes.

jakebeal commented 5 years ago

Another example of an insulator-like object is in this paper of mine: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0176013, which involves padding the 5' side of promoters to remove the effects of upstream sequence on them.

Do you think we should try to find a new term, or ask SO to adjust the definition of the current term? I'd tend towards the latter, personally.

shyambhakta commented 5 years ago

Actually, I feel our definition of insulator is a quite open-ended intent-driven engineering-specific term that describes sequences that facilitate reliable part reuse. Whereas the current SO insulator definition already has a longstanding distinct functional genetics definition that they perhaps ought not change. https://en.wikipedia.org/wiki/Insulator_(genetics)

Any way we can ask for a new SO term, something like "engineered gene expression insulator"? I considered that such could be a parent term for the current insulator, but I can't resolve the fact that the classical chromatin domain insulator isn't something that is defined by intent, and the only way such a broad insulator definition as ours can make sense is in the synthetic, engineering sense.

JS3xton commented 5 years ago

I feel like we should formalize an Inert DNA Spacer glyph, ask for an Inert DNA Spacer SO term, and abandon a generic Insulator glyph and its associated Insulator SO term.

Inert DNA Spacers, comprising inert DNA upstream and/or downstream of a promoter, have been used in multiple studies at this point, including:

Their illustration has been inconsistent, however:

Hatched box:

Did not illustrate.
Open circle with an X:

Black bar:

Concentric boxes Insulator glyph:

I tend to favor (5) or (3), but this is something that could be put to a vote.

Regarding other insulators, I believe the ones Shyam listed all have better, more specific glyphs, rendering a generic insulator glyph unnecessary.

shyambhakta commented 5 years ago

(1) and (4), hatched and black boxes, draw a lot of attention to a feature that is intended to be, well, passive/"inert". Even if it can be tuned to change a downstream promoter's activity, doing so with intent would better classify the "inert DNA" as an actual part of the promoter.

I like (3), ⊗, because it won't overlap with the existing insulator glyph (5), which I think is important if "inert DNA spacer" is supposed to be a subset of the generic insulator.

In an unresolved issue, I was discussing the problem of rotationally-symmetric glyphs like restriction sites and the origin not being explicit in their direction. I wonder if there's a way to prevent this glyph from having the same problem. Maybe make it a box instead of circle containing an × sitting on top of the line. Also, I've seen the –⊗– glyph used to denote terminators in many iGEM projects, even up to last year, so it might be misconstrued in the beginning.

P.S. This brings me to ask SBOL folks in Boston, maybe @jakebeal, to organize an actually useful iGEM workshop: one that teaches students how to use SBOL visual at least. I remember in the past years, no workshop I've been to was useful or didactic (or would have been even if I were a neophyte) and in cases, they were ridiculous. But an SBOL workshop could set a useful standard in teams.

JS3xton commented 5 years ago

classify the "inert DNA" as an actual part of the promoter

I actually think this should be done (as you can see in (5)).
I appreciate avoiding rotational symmetry, and I initially liked (6). It kinda resembles the Engineered Region glyph, though (rectangle on top of the DNA line), which may be undesirable.

jakebeal commented 4 years ago

Reviving this thread, I think the next step here is to get an appropriate SO term added, and so have opened an issue to do so: https://github.com/The-Sequence-Ontology/SO-Ontologies/issues/471

jakebeal commented 4 years ago

We've now got the SO term: SO:0002223

id: SO:0002223
name: inert_DNA_spacer
def: "Sequences that decrease regulatory interactions between a promoter, its 5' context and/or the translational unit(s) it regulated. Spacers can affect both translation and transcription." [PMID:20843779, PMID:24933158, PMID:27034378, PMID:28422998]
synonym: "Inert DNA Spacer" RELATED []
xref: BioRXiv:https\://doi.org/10.1101/584664 "https://www.biorxiv.org/content/10.1101/584664v1"
is_a: SO:0005836 ! regulatory_region
created_by: david
creation_date: 2019-09-06T19:05:52Z

Based on this, would somebody like to put together the pull request?

jakebeal commented 4 years ago

I'd like to move forward on this one, and the one question is whether we want to go with symmetric circle-x (3) or asymmetric box-x (6).

I am leaning toward the symmetric circle, for the same reason as @JS3xton. It does have the potential for confusion with ORI, but if it's RECOMMENDED size is smaller, I don't think that will be a big issue.

Others thoughts?

JS3xton commented 4 years ago

Yeah, neither option is perfect, but I also lean towards symmetric circle-x (3).

shyambhakta commented 4 years ago

I'm fine with either. –⊗– is perhaps the one to propose. Inert spacers don't have a functional directionality if they're truly inert, anyways.

graik commented 4 years ago

Well, this one has a directionality, doesn't it? The sequence only works on one direction and only in one frame.

Greetings Raik

On Thu, May 7, 2020, 08:44 Shyam Bhakta notifications@github.com wrote:

I'm fine with either. –⊗– is perhaps the one to propose. Inert spacers don't have a functional directionality if they're truly inert, anyways.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/SynBioDex/SBOL-visual/issues/34#issuecomment-625041470, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAOGZXLIXXQMHEOAKCZMFRTRQJDFJANCNFSM4ENSXX5A .

shyambhakta commented 4 years ago

@graik Are you referring to one of the literature examples above? None I see have directionality or frame. ––––––––––––––––––––––––––––– inert_DNA_spacer SO definition: Sequences that decrease regulatory interactions between a promoter, its 5' context and/or the translational unit(s) it regulated. Spacers can affect both translation and transcription [PMID:20843779, PMID:24933158, PMID:27034378, PMID:28422998]

I just realized, @jakebeal @JS3xton , that this definition wouldn't include examples comprising, e.g.,

insulator between promoter and 5′ UTR (technically an addition to the UTR). In designs of the BioDesigner strain selection/counterselection cassette, I remember Rob Egbert having used UNSs before and after promoters and after terminators, even, just to insulate from genetic and genomic context (at least so was is intent).
low-design-criteria sequence to space out distal operator from proximal ones in a synthetic promoter (for looping of the intervening DNA for added repression). In this paper from my lab, Chen Ye spaces out the distal and proximal operators (lac operators (lacO) in the linked fig), using ~350 nt segment of the completely unrelated human gene MYC, which acts as an inert DNA spacer by virtue of literally spacing the functional elements apart, no "regulatory insulation."
In the same paper, he uses a random ~1 kb chunk of the T7 RNAP CDS as an inert spacer upstream of all such promoters, so that genes don't get pulled into a promoter loop, or that loops aren't disrupted by upstream transcription bubbles. Perhaps this can still be described as insulation from promoter 5′ context.
I can imagine spacing any nondistinct parts (whatever they may be) apart with inert spacers so that when one functional part varies, just by virtue of the spacer's length and constancy, the other part's function is better insulated, because the local DNA context is more that of the constant inert spacer than the more distant varying part.

Sorry I always write so much, but do you get my drift? Perhaps even the new SO term is too specific for existing sequence that ought to qualify as inert DNA spacers.

A possible definition, if it is even possible to revise the term: Low-sequence-constraint DNA that isn't intended to specifically encode information or an active function. That's not to say its sequence cannot impact function, e.g. the AT-richness in a promoter's UP element region and discriminator, or cryptic promoter motifs in the aforementioned MYC gene spacer. The point is that the spacer is not intended to have specific information in its sequence.

JS3xton commented 4 years ago

I agree with @shyambhakta and think the definition of inert_DNA_spacer is too specific.

My attempt at a definition: DNA designed to have no biological activity.

graik commented 4 years ago

Sorry, wrong thread... please ignore my comment :/

On Thu, May 7, 2020 at 10:04 AM Raik Grünberg raik.gruenberg@gmail.com wrote:

Well, this one has a directionality, doesn't it? The sequence only works on one direction and only in one frame.

Greetings Raik

On Thu, May 7, 2020, 08:44 Shyam Bhakta notifications@github.com wrote:

I'm fine with either. –⊗– is perhaps the one to propose. Inert spacers don't have a functional directionality if they're truly inert, anyways.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/SynBioDex/SBOL-visual/issues/34#issuecomment-625041470, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAOGZXLIXXQMHEOAKCZMFRTRQJDFJANCNFSM4ENSXX5A .

--

Raik Grünberg http://www.raiks.de/contact.html

jakebeal commented 3 years ago

I've requested a generalization of the term to cover any unwanted interaction: https://github.com/The-Sequence-Ontology/SO-Ontologies/issues/494

On the assumption that will resolve (though it may take a bit), I have also set up an SEP for adding the circle-X Inert DNA Spacer glyph and deprecating the old insulator glyph that nobody really uses: #103.

I am going to close this issue in favor of the new SEP issue, where we can continue discussion as needed.

SynBioDex / SBOL-visual

Insulator is too biologically specific #34