jakebeal
commented
3 years ago There is a very similar issue for the Promoter glyph. At one extreme, promoter is sometimes used more abstractly to refer to the whole region, including regulatory binding sites, spacer material, TATA box / -35 site, and transcriptional start site. At the opposite extreme, I have seen promoter sometimes used to indicate precisely and only the TSS. And, of course, there are various levels of detail in between, when somebody wants to talk about some of these aspects but not others.
In the case of promoters, SO has at least partially resolved this for us, because it has defined that, for example, the -35 site (http://www.sequenceontology.org/browser/current_release/term/SO:0000176) is a child of Promoter (http://www.sequenceontology.org/miso/current_release/term/SO:0000167). The basic problem is this there, however.
As I think about this, I find that for myself at least, the big divide is between transcripts and things that control transcription. The reason is that this tends to be a critical point when qualitatively changing the function of a biological system. Thus, the most core glyphs being used are:
- A thing that controls transcription: Promoter
- A thing that gets transcribed: Coding Sequence --> generalize to Transcript (SO:0000673)?
- A thing that includes both: Engineered region --> generalize to Region (SO:0000001)?
This would put the resistance genes using the rectangle for Region and the polycistronic cytochrome using a pentagon for Transcript. This would also be consistent with the recent work on introns (#63) and 2A regions (#78).
Some additional notes:
- Engineered Region shouldn't be generalized all the way to Sequence Feature (SO:0000110) because that includes zero-length elements and also because that is used for the unspecified glyph, to denote things that are missing information.
- Transcript would mean the pentagon also covers ncRNA, which has its own glyph. I think that may be OK, however, since choosing to use a more specific glyph is always allowed.
toddslaby
rsc3
Gonza10V
fxbuson
shyambhakta
It has historically been common to simplify the depiction of an entire gene, operon, or gene cluster as a single glyph that is equivalent in form to that of the current SBOL CDS glyph.
Wild-type resistance genes, e.g., Kanᴿ and Chlᴿ (kanamycin and chloramphenicol resistance) are labeled with the phenotype imparted, but popularly shown with just a CDS glyph (Fig A). The inclusion of more detailed promoter, RBS, and terminator glyphs makes for undesired clutter, and I find it very rare that these control elements are annotated precisely even in sequence files. As resistance is a phenotype, it can't technically be the label for just a CDS, which would instead require the CDS names, e.g., aphA1/nptI/aphA2/nptII or cat for these resistance genes. Many possible CDSs may encode proteins that confer the same resistance phenotype, but because these CDS names are less familiar, often not even known by the users, it's often preferred to write the more practical resistance phenotype.
This concept often extends to transcription factor genes, e.g. lacIq which is a stronger promoter mutant of the wild-type lacI gene, and thus cannot describe just a CDS (Fig A). Another case is a constitutive arabinose transcriptional activator + transporter operon araCE, which may be represented by just a single CDS glyph labeled ara or araCE. It's not clear that an SBOL CDS glyph can be used more broadly for such whole genes.
The concept also often extends to larger operons and gene clusters, e.g., native cytochrome c maturation gene cluster cloned into a plasmid under synthetic control (Fig A). But instead of taking a bunch of space with eight CDSs in order labeled ccmA ccmB ccmC ccmD ccmE ccmF ccmG ccmH, one may want to simplify it to the equivalent of a single CDS glyph labeled ccm or ccmA–H or ccmABCDEFGH, which represents the whole translational unit, or possibly the whole operon / gene cluster, from promoter to terminator, especially when using the native control elements. Another example: sets of plasmid partitioning genes used to stabilize plasmid inheritance are often named with the prefix par, and their operon(s) (e.g. parABCD) and their associated promoters/terminators are often simplified to the equivalent of just one CDS glyph labelled par, in the primary direction of transcription if there is one (Fig A). Users see it and know "this is the partitioning locus" without caring about the finer details.
Even more abstractly, the CDS glyph is used to represent a genetic locus, with the direction of the arrow representing the direction of transcription of a key gene, as in the sole replicator protein in many replication origins (Fig B–D), or perhaps the direction of DNA replication if unidirectional from an origin, or even an arbitrary direction, which is at least useful in establishing directionality of the corresponding defined sequence, something that a rotationally-symmetric origin glyph "○" doesn't capture. For any arbitrary locus, one ought to be able to refer to nucleotide indices within it, with a corresponding SBOL glyph making evident where the beginning and end is, and the glyph length ought to be able to be scaled for longer and shorter loci, things that SBOL glyphs ought to have options for in representing, for example, a replication origin or future glyphs for other specific kinds of genetic loci. If I were to say the helicase-binding box of pSC101 is within the first 100 nt of the sequence, Fig B doesn't make it clear relatively where that is, whereas Fig D does and Fig C would if it were described as in the oriV.
So what can we decide is the recommendation for a glyph of a whole gene (single gene or operon), and/or an entire gene cluster / "locus" with all its tx/tl initiation/termination elements? Some people write it in a box instead of an arrow, but this isn't a general solution as gene direction is still useful to describe in many cases, hence the other traditional use of the CDS arrow/pentagon. I'm not sure such genes necessary qualify as "engineered regions" to be able to use a box; they're often natural sequences used as a whole for a phenotype, part of wanting to abstract into a single glyph in the same way the many genes with distinct promoters, CDSs, ncRNAs, etc. mediating replication_origin function are abstracted into a single origin glyph ○.
The specs of the CDS glyph shows association with only CDS SO:0000316: A contiguous sequence which begins with, and includes, a start codon and ends with, and includes, a stop codon. The far broader term "gene" is the proposition for what the popular usage of the CDS glyph is also describing in the earlier examples, gene SO:0000704 A region (or regions) that includes all of the sequence elements necessary to encode a functional transcript. A gene may include regulatory regions, transcribed regions and/or other functional sequence regions. Its children include gene_with_polycistronic_transcript and gene_cassette, and so cover some of the examples above. However, the usage encompassing multiple genes in multiple operons/directions would require another level up, gene_group SO:0005855**, A collection of related genes.
A dual-use of the CDS glyph for both a real start-stop codon CDS and entire gene or gene_group is seemingly contradictory, and I'd expect contention on the matter if the consensus is to allow it. And if not, the design of a new gene / gene_group glyph could also be controversial since it violates a very old tradition, which might suggest that a new glyph would have to be at least quite similar to the CDS glyph.