Trace the Biosynthetic Path of Glycans
trace_biosynthesis.RdReconstruct biosynthetic pathways for one or more glycans. The default
method uses known enzyme rules in a forward, multi-target breadth-first
search. To infer structure-driven paths without enzyme specificity, use
trace_biosynthesis_virtual().
Usage
trace_biosynthesis(
glycans,
enzymes = NULL,
max_steps = NULL,
filter = NULL,
max_virtual_steps = 0L
)Arguments
- glycans
A
glyrepr::glycan_structure()vector, or a character vector of strings supported byglyparse::auto_parse(). Can also be a single glycan. If multiple glycans are provided, the starting structure will be decided by the first glycan. Therefore, please make sureglycansare not of mixed glycan types.- enzymes
A character vector of concrete gene symbols or abstract activity names, or a list of
enzyme()objects. Concrete and abstract enzymes cannot be mixed. IfNULL(default), all available concrete enzymes will be used.- max_steps
Integer, maximum number of enzymatic steps to search, or
NULLto infer it from the largest target glycan. The inferred value is the maximum monosaccharide and substituent count plus 4 for N-glycans, and the maximum count minus 1 for other glycans.- filter
Optional function to filter generated glycans at each step. Should take a
glyrepr::glycan_structure()vector as input and return a logical vector of the same length. It filters generated products.- max_virtual_steps
Integer, maximum number of target-directed virtual enzyme steps allowed when no fully enzymatic path exists. Default is
0L, which disables virtual fallback. See the "Virtual fallback" section for more details.
Value
A glyenzy_biosynthesis_network object inheriting from
igraph::igraph() and representing the synthesis path(s). Vertices
represent glycan structures, with IUPAC-condensed strings in the name
attribute and a logical target attribute indicating whether each vertex
is a target glycan. At most one directed edge connects each substrate and
product. Every edge has an integer step, a logical is_virtual, a
scalar display label in enzyme, and a list-valued enzymes attribute
containing concrete enzyme candidates. For known reactions, enzyme
combines the candidates with " / "; for virtual reactions, it contains
the structural virtual-enzyme name and enzymes is empty unless concrete
candidates were annotated.
For multiple targets, the graph includes all synthesis paths needed to reach every target glycan.
Input compatibility
trace_biosynthesis(), trace_biosynthesis_virtual(),
path_biosynthesis(), and path_biosynthesis_virtual() accept only
glycans that share one monosaccharide type ("concrete" or "generic")
and one structure level ("intact" or "topological"). Structures with
mixed generic and concrete residues, partial structures, and missing
structures are not supported. In the path_*() functions, from and to
must have the same monosaccharide type and structure level.
Use glyrepr::convert_to_generic() to standardize monosaccharide types or
glyrepr::remove_linkages() to standardize structures at the topological
level.
Virtual fallback
Sometimes the biosynthesis network of a glycan cannot be fully resolved; i.e., some enzymatic steps are not inferred to be catalyzed by any known enzyme ("bad" steps). By default, an error is raised for these glycans.
max_virtual_steps provides a fallback for these glycans.
For a "bad" step, a virtual enzyme is assigned to allow the algorithm to
continue. For example, for the O-GalNAc core 5
"GalNAc(a1-3)GalNAc(a1-", an "a3GalNAcT" is assigned to the step that adds
the a3 GalNAc. Unsupported sulfate additions similarly use "3SulfoT",
"6SulfoT", or "?SulfoT".
Therefore, max_virtual_steps can also be interpreted as
"the maximum number of glycosidic bonds or sulfate transfers that cannot be
assigned by a known enzyme."
Increasing this number loosens the criteria.
Important notes
Here are some important notes for all functions in the glyenzy package.
Applicability
Known-enzyme algorithms and enzyme information in glyenzy are applicable only to humans. Curated coverage is strongest for N-glycans and O-glycans and also includes selected glycosphingolipid headgroups and other glycan contexts. Lipid and protein aglycones are not represented, so glycolipid rules model the carbohydrate headgroup with ceramide omitted. Results may be inaccurate for unsupported glycan contexts or other species (e.g., plants, insects).
Inclusiveness
The algorithm takes an intentionally inclusive approach, assuming that all possible isoenzymes capable of catalyzing a given reaction may be involved. Therefore, results should be interpreted with caution.
For example, in humans, detection of the motif "Neu5Ac(a2-3)Gal(b1-" will return both "ST3GAL3" and "ST3GAL4". In reality, only one of them might be active, depending on factors such as tissue specificity.
Concrete glycans by default
Most functions only work for glycans containing concrete residues
(e.g., "Glc", "GalNAc"), and not for glycans with generic
residues (e.g., "Hex", "HexNAc").
Inputs with generic residues are supported where explicitly
documented, such as trace_biosynthesis() and path_biosynthesis().
Substituents
Sulfate substituents are supported. Other substituents, such as
phosphorylation and methylation, are not supported. Use
glyrepr::remove_substituents() when unsupported substituents are present.
Incomplete or non-concrete glycan structures
Some functions accept glycan structures that are incomplete, partially
degraded, or contain generic residues. When accepted, glycans with a
glyrepr::get_structure_level() other than "intact", or with a
glyrepr::get_mono_type() other than "concrete", are matched with the
lenient motif matching mode in glymotif. A warning is raised because enzyme
predictions may be less reliable.
Starting points
For known-enzyme path inference:
For N-glycans, the starting structure is assumed to be "Glc(3)Man(9)GlcNAc(2)", the N-glycan precursor transferred to Asn by OST.
For O-GalNAc glycans, the starting structure is assumed to be "GalNAc(a1-".
For O-GlcNAc glycans, the starting structure is assumed to be "GlcNAc(b1-".
For O-Man glycans, the starting structure is assumed to be "Man(a1-".
For O-Fuc glycans, the starting structure is assumed to be "Fuc(a1-".
For O-Glc glycans, the starting structure is assumed to be "Glc(b1-".
For GlcCer glycans, the starting structure is assumed to be "Glc(b1-",
For GalCer glycans, the starting structure is assumed to be "Gal(b1-"
Examples
library(glyrepr)
library(glyparse)
# Rebuild the biosynthetic pathway of a single glycan
glycan <- "Neu5Ac(a2-3)Gal(b1-4)[Fuc(a1-3)]GlcNAc(b1-3)Gal(b1-3)GalNAc(a1-"
path <- trace_biosynthesis(glycan)
# Rebuild pathways for multiple glycans
glycans <- c(
"Neu5Ac(a2-3)Gal(b1-4)[Fuc(a1-3)]GlcNAc(b1-3)Gal(b1-3)GalNAc(a1-",
"Gal(b1-4)[Fuc(a1-3)]GlcNAc(b1-3)Gal(b1-3)GalNAc(a1-"
)
path <- trace_biosynthesis(glycans)
# View the path
igraph::as_data_frame(path, what = "edges")
#> from
#> 1 GalNAc(a1-
#> 2 Gal(b1-3)GalNAc(a1-
#> 3 GlcNAc(b1-3)Gal(b1-3)GalNAc(a1-
#> 4 Gal(b1-4)GlcNAc(b1-3)Gal(b1-3)GalNAc(a1-
#> 5 Gal(b1-4)GlcNAc(b1-3)Gal(b1-3)GalNAc(a1-
#> 6 Neu5Ac(a2-3)Gal(b1-4)GlcNAc(b1-3)Gal(b1-3)GalNAc(a1-
#> to
#> 1 Gal(b1-3)GalNAc(a1-
#> 2 GlcNAc(b1-3)Gal(b1-3)GalNAc(a1-
#> 3 Gal(b1-4)GlcNAc(b1-3)Gal(b1-3)GalNAc(a1-
#> 4 Fuc(a1-3)[Gal(b1-4)]GlcNAc(b1-3)Gal(b1-3)GalNAc(a1-
#> 5 Neu5Ac(a2-3)Gal(b1-4)GlcNAc(b1-3)Gal(b1-3)GalNAc(a1-
#> 6 Neu5Ac(a2-3)Gal(b1-4)[Fuc(a1-3)]GlcNAc(b1-3)Gal(b1-3)GalNAc(a1-
#> enzyme is_virtual step enzymes
#> 1 C1GALT1 FALSE 1 C1GALT1
#> 2 B3GNT3 FALSE 2 B3GNT3
#> 3 B4GALT1 / B4GALT2 / B4GALT3 / B4GALT4 / B4GALT5 FALSE 3 B4GALT1,....
#> 4 FUT3 / FUT4 / FUT5 / FUT6 / FUT9 FALSE 4 FUT3, FU....
#> 5 ST3GAL4 / ST3GAL6 FALSE 4 ST3GAL4,....
#> 6 FUT5 / FUT6 / FUT7 FALSE 5 FUT5, FU....