Package 'glyenzy'

Description

Return a separate collection of simplified N-glycan reaction rules. These enzymes represent reaction activities rather than individual genes. They are never included in db_enzymes(), enzymes_from_rnaseq(), or the default enzyme sets of other functions. Individual activities can also be retrieved by name with enzyme(), for example enzyme("ManII").

Usage

abstract_enzymes(return_str = FALSE)

Arguments

Details

The collection contains ManI, GnTI, ManII, GnTII, GnTIII, a6FucT, GnTIV, GnTV, b4GalT, iGnT, a3SiaT, and GlcH: 12 enzymes with 32 reaction rules. All have glycan_type = "N" and species = "unspecified". The additional localization field records "cis", "medial", "trans", or "ER"; it is descriptive and does not impose compartment order during tracing.

The trimming activities follow the corresponding concrete enzymes in the ordinary database. ManI combines the 16 distinct whole-glycan rules of MAN1B1, MAN1A1, MAN1A2, and MAN1C1. ManII uses the shared MAN2A1/MAN2A2 rule, including its core alignment and rejection of bisecting GlcNAc or a galactosylated alpha1-3-arm GlcNAc. It removes either the alpha1-3 or alpha1-6 mannose first, preserving both Man4 intermediates. These rules give the same trimming reactions and network topology as their concrete counterparts; enzyme labels remain abstract.

Core transferase rules also follow the concrete database, including their acceptors, alignment, and rejects: GnTI uses MGAT1; GnTII uses MGAT2; GnTIII uses MGAT3; a6FucT uses FUT8; GnTIV uses MGAT4A/MGAT4B; and GnTV uses the N-glycan rule shared by MGAT5/MGAT5B. The O-mannose-specific MGAT5B rule is not included. GnTIII adds bisecting GlcNAc and has localization = "medial" in this abstract model.

FUT8, MGAT1, and MGAT2 have no explicit rejects in the concrete database; their acceptor and alignment restrictions are retained instead of adding new exclusions. GnTIII rejects galactosylated alpha1-3-arm antennae. GnTIV and GnTV retain their bisecting-GlcNAc and antenna-specific rejects. A reject excludes the matching reaction site, not every site on a glycan.

iGnT rejects the first LacNAc extension on both the beta1-2 and beta1-4 antennae of the alpha1-3 arm, without excluding sites on another arm. Because only iGnT initiates LacNAc extension in this collection, that arm remains unextended in networks generated from the default precursor. The B3GNT4 reject for terminal Gal-GlcNAc-Gal-GlcNAc additionally prevents a second LacNAc extension. Thus the alpha1-6 arm can extend once, but not repeatedly through this activity.

a3SiaT retains the ST3GAL4/ST3GAL6 reject for alpha1-3 fucose on its Gal-GlcNAc acceptor. b4GalT has no explicit reject in the corresponding N-glycan B4GALT rules. Only restrictions for the represented N-glycan activity are imported; unrelated O-glycan and glycolipid reactions are excluded. These terminal activities retain the table acceptors, so this collection is not a replacement for the complete concrete database.

Ambiguous linkages use the existing motif-matching semantics, including potentially compatible reject matches. Product-motif statistics do not apply substrate rejects and do not establish that a product is reachable.

GlcH combines the three glucose-removal steps represented by MOGS and GANAB in the ordinary database. It removes one residue per step from the complete Glc3Man9GlcNAc2 precursor, retaining Glc2 and Glc1 intermediates. Thus the collection can trace from the usual trace_biosynthesis() starting structure without adding ordinary enzymes or changing the start.

Pass activities explicitly, for example enzyme("ManI") to apply_enzyme(), or activity names or the whole list from abstract_enzymes() to trace_biosynthesis(). Abstract reactions are explicit enzyme rules and have is_virtual = FALSE in biosynthesis networks. Existing restrictions on GH involvement statistics and residue matching still apply; this collection does not add GH support to those interfaces.

Value

A named list of glyenzy_abstract_enzyme objects, or a character vector when return_str = TRUE. Each object also inherits from glyenzy_gt_enzyme or glyenzy_gh_enzyme, and glyenzy_enzyme.

References

The simplified Golgi rules follow the user-supplied reaction table based on the N-glycosylation model of Spahn et al. (2016), with branch notation described by Liang et al. (2020). The supplied table takes precedence over differences in published versions, except for the concrete trimming and core-transferase rules, added GnTIII activity, and terminal-activity rejects described above. See the bundled rule-data attribution.

Spahn et al. (2016). A Markov chain model for N-linked protein glycosylation. doi:10.1016/j.ymben.2015.10.007.

Liang et al. (2020). A Markov model of glycosylation elucidates isozyme specificity and glycosyltransferase interactions for glycoengineering. doi:10.1016/j.crbiot.2020.01.001.

Examples

abstract_enzymes(return_str = TRUE)
enzyme("ManII")

man5 <- paste0(
  "Man(a1-3)[Man(a1-3)[Man(a1-6)]Man(a1-6)]",
  "Man(b1-4)GlcNAc(b1-4)GlcNAc(b1-"
)
apply_enzyme(man5, abstract_enzymes()[["GnTI"]])
trace_biosynthesis(man5, enzymes = abstract_enzymes(), max_virtual_steps = 0L)

Description

This function simulates the action of an enzyme on a glycan. It returns all possible products generated by the enzyme with the given glycans.

Usage

apply_enzyme(glycans, enzyme, return_list = NULL, structure_level = "intact")

Arguments

Value

A glyrepr::glycan_structure() vector, or a list of such vectors.

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)

# Use `glycan_structure()` and `enzyme()`
glycan <- auto_parse("GlcNAc(b1-2)Man(a1-3)[Man(a1-6)]Man(b1-4)GlcNAc(b1-4)GlcNAc(b1-")
apply_enzyme(glycan, enzyme("MGAT3"))

# Or use characters directly
apply_enzyme("GlcNAc(b1-2)Man(a1-3)[Man(a1-6)]Man(b1-4)GlcNAc(b1-4)GlcNAc(b1-", "MGAT3")

# Vectorized input
glycans <- c(
  "GlcNAc(b1-2)Man(a1-3)[Man(a1-6)]Man(b1-4)GlcNAc(b1-4)GlcNAc(b1-",
  "GlcNAc(b1-2)Man(a1-3)[GlcNAc(b1-2)Man(a1-6)]Man(b1-4)GlcNAc(b1-4)GlcNAc(b1-"
)
apply_enzyme(glycans, "MGAT3")

Description

autoplot.glyenzy_biosynthesis_network() draws the typed networks returned by trace_biosynthesis(), trace_biosynthesis_virtual(), path_biosynthesis(), and path_biosynthesis_virtual(). A layered DAG layout determines the node positions while accounting for converging biosynthesis routes. Candidate enzyme names stored on each reaction edge are combined into one label.

Usage

## S3 method for class 'glyenzy_biosynthesis_network'
autoplot(
  object,
  show_enzyme = TRUE,
  size = 0.4,
  node_gap = 0.25,
  level_gap = 0.6,
  width = 6,
  height = 6,
  units = c("in", "cm", "mm"),
  show_linkage = FALSE,
  orient = c("left", "right", "up", "down"),
  color_edge = FALSE,
  enzyme_label_style = c("condensed", "full"),
  highlight_target = NULL,
  ...
)

Arguments

Details

Glycan dimensions are measured before the network is laid out. Nodes at the same rank are separated by their rendered bounds, while rectangular edge caps stop arrows outside the source and target glycans. The plot uses a fixed-size panel inside a requested figure canvas so these clearances remain physical rather than changing with the coordinate range.

Concrete enzyme reactions use solid edge shafts, while virtual enzyme reactions use dashed shafts; arrowheads remain solid in both cases. This convention is documented rather than repeated in an in-plot legend.

Value

A ggraph/ggplot object with a fixed-size, collision-aware layered panel centered in the requested figure dimensions.

Figure size

width, height, and units describe the complete base figure returned by this method. Use the same dimensions with ggplot2::ggsave() or the corresponding knitr/Quarto figure options. Titles, legends, or margins added after this method returns may require additional output space.

Examples

## Not run: 
network <- trace_biosynthesis_virtual(
  "GlcNAc(b1-4)Gal(b1-3)GalNAc(a1-"
)
network_plot <- ggplot2::autoplot(network, width = 8, height = 5)
ggplot2::ggsave(
  "biosynthesis-network.png",
  network_plot,
  width = 8,
  height = 5,
  units = "in"
)

## End(Not run)

Description

Count how many times an enzyme is involved in the biosynthesis of a glycan.

Usage

count_enzyme(glycans, enzyme, method = c("motif", "path"))

Arguments

Value

An integer vector of the same length as glycans.

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)

# Use `glycan_structure()` and `enzyme()`
glycan <- auto_parse("Neu5Ac(a2-6)Gal(b1-4)GlcNAc(b1-")
count_enzyme(glycan, enzyme("ST6GAL1"))

# Or use characters directly
count_enzyme("Neu5Ac(a2-6)Gal(b1-4)GlcNAc(b1-", "ST6GAL1")

# Vectorized input
glycans <- c(
  "Neu5Ac(a2-6)Gal(b1-4)GlcNAc(b1-",
  "Gal(b1-4)GlcNAc(b1-"
)
count_enzyme(glycans, "ST6GAL1")

# Use reconstructed biosynthesis paths
count_enzyme(glycans, "ST6GAL1", method = "path")

Description

Return a named list of ordinary built-in enzymes, or a character vector of gene symbols if return_str is TRUE. Abstract activities are available separately through abstract_enzymes().

Usage

db_enzymes(
  return_str = FALSE,
  include_starter_gt = TRUE,
  include_npre_gt = TRUE
)

Arguments

Value

A list of enzyme()s or a character vector.

Examples

db_enzymes(return_str = TRUE)

Description

Use enzyme() with a gene symbol or an abstract name to load a predefined enzyme. For example, use enzyme("ST3GAL3") to load the enzyme ST3GAL3, and enzyme("ManII") for a general ManII (possibly MAN2A1 or MAN2A2). It returns an enzyme object that can be passed to any enzyme argument of all functions in this package. As a shortcut, passing a name is also valid, e.g. have_enzyme(glycan, "ST3GAL3").

Usage

enzyme(symbol)

Arguments

Value

A glyenzy_enzyme object. Abstract activities additionally inherit from glyenzy_abstract_enzyme.

Enzyme types (biological)

Enzymes can be classified by their biological rules. Three types of enzymes can be represented: glycosyltransferases (GTs), glycoside hydrolases (GHs), and sulfotransferases (STs).

• GTs catalyze the transfer of a sugar residue from a donor to an acceptor, thus building up glycan structures.

• GHs catalyze the removal of a sugar residue from a substrate, thus breaking down glycan structures.

• STs add a sulfate substituent to an existing glycan residue.

One special subtype of GTs are starter GTs (or initiating GTs), which catalyze the addition of the first sugar residue onto a non-glycan substrate, thus initiating glycosylation.

Enzyme types (computational)

There're also three types of enzymes you'll encounter in this package depending on how we define them, they are concrete enzymes, abstract enzymes, and virtual enzymes.

Concrete enzymes are the enzymes represented by gene symbols, e.g. B4GALT1. There can be many isoenzymes on this level. For example, B4GALT1/2/3/4/5 are all isoenzymes with similar substrate specificity.

Abstract enzymes are conventional enzymes without considering isoenzymes. For example, the above enzymes are described as b4GalT. Other examples include ManI, ManII, GnTI, etc. Currently only abstract enzymes for N-glycan biosynthesis are supported.

Both concrete and abstract enzymes can be used in enzyme or enzymes arguments throughout this package. A single enzymes collection must contain only concrete enzymes or only abstract enzymes; the two types cannot be mixed. Both types are also supported by the enzyme() function.

Another type of enzymes is called the virtual enzymes. You'll see them in the results of path_biosynthesis_virtual() and trace_biosynthesis_virtual(). Find out more in the documentation of these two functions.

Explanation about glyenzy_enzyme

An enzyme() is a list with the following elements:

1. name: the name of the enzyme, usually the gene symbol.

2. rules: a list of glyenzy_enzyme_rule objects. Each rule is a list with the following fields:

   • acceptor: the motif that the enzyme recognizes

   • acceptor_alignment: the alignment of the acceptor

   • rejects: the motifs that the enzyme rejects to act on

   • requires: optional OR-valued context motifs. When present, at least one requirement must match.

   • product: the product generated by the enzyme

   • acceptor_idx: the node index of the acceptor where the enzyme acts on. For GTs, this is the node new residue is attached to. For GHs, this is the node that is removed. For starter GTs, acceptor_idx is 0.

   • product_idx: the node index of the product residue in the product structure. For GTs, this is the index of the newly added residue in the product. For STs, this is the product node corresponding to the modified acceptor residue. For GHs, this is NULL.

   • new_residue: the new residue added by a GT. For GHs and STs, this is NULL.

   • new_linkage: the linkage of the new residue added by a GT. For GHs and STs, this is NULL.

   • new_substituent: for STs, the sulfate token added by the rule, such as "6S". This field is absent for other enzyme types.

   • product_sub: for STs, the canonical complete substituent state of the modified product residue, such as "3S,6S". This field is absent for other enzyme types.

3. type: the broad type of the enzyme, "GT" for glycosyltransferase, "GH" for glycoside hydrolase, or "ST" for sulfotransferase. Starter GTs are encoded as type = "GT".

4. species: the species of the enzyme, e.g. "human" or "mouse".

5. glycan_type: NULL for enzymes that can act on all glycan classes, or one or more of "N", "O", "lipid", and "free" for class-specific enzymes. Structures ending in Glc or Gal without a represented carrier are compatible with both "lipid" and "free".

You can see all these information by printing the enzyme object.

Examples

library(glyrepr)

enzyme("ST3GAL3")
enzyme("ManII")

Description

enzymes_from_rnaseq() selects built-in enzymes whose genes are expressed at or above a TPM threshold. When mat contains multiple columns, the mean TPM across columns is used for each gene.

Usage

enzymes_from_rnaseq(mat, threshold = 1)

Arguments

Value

A named list of enzyme() objects. Genes not represented in the built-in enzyme database are ignored.

References

Huang, Y.-F., Aoki, K., Akase, S., et al. (2021). Global mapping of glycosylation pathways in human-derived cells. Developmental Cell, 56, 1195-1209. doi:10.1016/j.devcel.2021.02.023.

Examples

tpm <- matrix(
  c(0.2, 1, 3),
  ncol = 1,
  dimnames = list(c("FUT8", "ST3GAL3", "MOGS"), "sample")
)
enzymes_from_rnaseq(tpm)

Description

This function returns all possible isoenzymes associated with the biosynthetic steps of the input glycan. Note that this function ignores the residues in glycans that cannot be matched to any enzyme rules.

Usage

find_enzyme(glycans, return_list = NULL, method = c("motif", "path"))

Arguments

Value

A character vector or a list of character vectors (see return_list parameter), each containing the names of enzymes involved in the biosynthesis of the corresponding glycan.

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)

# Use `glycan_structure()`
glycans <- auto_parse(c(
  "GlcNAc(b1-2)Man(a1-3)[Man(a1-6)]Man(b1-4)GlcNAc(b1-4)GlcNAc(b1-",
  "GlcNAc(b1-2)Man(a1-3)[GlcNAc(b1-2)Man(a1-6)]Man(b1-4)GlcNAc(b1-4)GlcNAc(b1-"
))
find_enzyme(glycans)

# Or use characters directly
find_enzyme("GlcNAc(b1-2)Man(a1-3)[Man(a1-6)]Man(b1-4)GlcNAc(b1-4)GlcNAc(b1-")

# Use reconstructed biosynthesis paths
find_enzyme(glycans, method = "path")

Description

This function simulates the action of enzymes on glycans. Think of it like a primordial soup where you put in a few glycans and enzymes, and let them interact to generate new glycans.

grow_glycans_step() performs one round of enzyme action, while grow_glycans() performs multiple rounds. The only difference between grow_glycans_step() and grow_glycans(n_steps = 1) is that the latter returns the original input glycans as well. For both, a vector of unique glycan structures is returned.

The number of glycans generated by grow_glycans() is typically exponential, and can quickly become very large. Therefore, it is recommended to use a small number of steps and carefully select the enzymes. Also, you can use the filter argument to prune the results after each round.

Usage

grow_glycans_step(glycans, enzymes)

grow_glycans(glycans, enzymes, n_steps = 5, filter = NULL)

Arguments

Value

A glyrepr::glycan_structure() vector of all unique glycans generated.

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

# Use `grow_glycans_step()` to build glycans step by step
glycan <- "GlcNAc(b1-2)Man(a1-3)[Man(a1-6)]Man(b1-4)GlcNAc(b1-4)GlcNAc(b1-"
glycan |>
  grow_glycans_step("MGAT2") |>
  grow_glycans_step("B4GALT1") |>
  grow_glycans_step("ST3GAL3")

# Use `grow_glycans()` to simulate a primordial soup
glycans <- c(
  "GlcNAc(b1-2)Man(a1-3)[Man(a1-6)]Man(b1-4)GlcNAc(b1-4)GlcNAc(b1-",
  "GlcNAc(b1-2)Man(a1-3)[GlcNAc(b1-2)Man(a1-6)]Man(b1-4)GlcNAc(b1-4)GlcNAc(b1-"
)
enzymes <- c("B4GALT1", "ST3GAL3")
grow_glycans(glycans, enzymes, n_steps = 5)

# Use `filter` to prune the results after each round
# Here we keep only glycans that are synthesized by MGAT2
grow_glycans(glycans, enzymes, n_steps = 5, filter = ~ have_enzyme(.x, "MGAT2"))

Description

Glycans are produced through a series of enzymatic reactions. This function checks whether a specific enzyme participates in the biosynthesis of a given glycan (or glycans).

Usage

have_enzyme(glycans, enzyme, method = c("motif", "path"))

Arguments

Value

A logical vector of the same length as glycans.

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-"

Algorithm

The overall approach is straightforward: for each reaction rule associated with the enzyme, the function checks whether the corresponding product motif appears in the glycan. If any rule matches, the function returns TRUE.

For N-glycans, additional logic is applied to handle special cases. Products of MGAT1 are often further trimmed by glycoside hydrolases, meaning that the final glycan product may no longer contain the original motif. In these cases, the function instead looks for specific motif markers to determine enzyme involvement.

Examples

library(glyrepr)
library(glyparse)

# Use `glycan_structure()` and `enzyme()`
glycan <- auto_parse("Neu5Ac(a2-6)Gal(b1-4)GlcNAc(b1-")
have_enzyme(glycan, enzyme("ST6GAL1"))

# Or use characters directly
have_enzyme("Neu5Ac(a2-6)Gal(b1-4)GlcNAc(b1-", "ST6GAL1")

# Vectorized input
glycans <- c(
  "Neu5Ac(a2-6)Gal(b1-4)GlcNAc(b1-",
  "Gal(b1-4)GlcNAc(b1-"
)
have_enzyme(glycans, "ST6GAL1")

# Use reconstructed biosynthesis paths
have_enzyme(glycans, "ST6GAL1", method = "path")

Description

This function creates a custom enzyme object, which can be used in all functions in this package that accept an enzyme argument. Most of the time, you can use enzyme() with a gene symbol to get a predefined enzyme, or just use the gene symbol directly in function arguments. Use this function only when you need to create an enzyme with custom rules. See the "Enzyme rules" section below for more details.

You can create regular glycosyltransferases (GTs), glycoside hydrolases (GHs), and sulfotransferases (STs) with this function. Special enzymes like starter GTs for glycosylation initiation are reserved for predefined enzymes.

Usage

make_enzyme(name, rules, type, species, glycan_type = NULL)

Arguments

Value

A glyenzy_enzyme object.

Enzyme rules

The enzyme rules are the core of an enzyme object. Each rule defines one possible reaction of an enzyme, and an enzyme can have multiple rules.

We now explain each field of an enzyme rule in detail.

acceptor

The acceptor is the glycan motif (substructure) that the enzyme recognizes. It can be as simple as a single residue, like ⁠Man(a1-⁠, or a more complex motif, like ⁠Man(a1-3)[Man(a1-6)]Man(b1-4)GlcNAc(b1-4)GlcNAc(b1-⁠. Note that the acceptor does not define where the enzyme acts on, which is determined collectively by the acceptor and product fields.

acceptor_alignment

Under the hood, the acceptor is detected by glymotif::match_motif(). To match a motif, we need to specify the alignment of the motif. Possible values are "substructure", "core", "terminal" and "whole". See glymotif::have_motif() for details.

rejects

Some enzymes are really picky. They will reject to act on certain glycans if some motifs are present, even if the acceptor is perfectly matched. For example, if we want to define an enzyme that adds a Neu5Ac residue to the Gal residue of a ⁠Gal(b1-4)GlcNAc(b1-⁠ motif, but only when there is no a1-3 Fuc on that GlcNAc, we can set "Gal(b1-4)GlcNAc(b1-" as the acceptor and "Fuc(a1-3)[Gal(b1-4)]GlcNAc(b1-" as the rejects. You can provide multiple rejects as a vector. For example, if we want to reject ⁠Gal(b1-4)GlcNAc(b1-⁠ with either a1-3 Fuc on GlcNAc or a1-2 Fuc on Gal, we can set c("Fuc(a1-3)[Gal(b1-4)]GlcNAc(b1-", "Fuc(a1-2)Gal(b1-4)GlcNAc(b1-") as the rejects.

Note that here are some restrictions on the rejects:

• When acceptor_alignment is "whole", rejects cannot be set.

• The acceptor must be the substructure of all rejects. For example, in the a1-3 Fuc example above, you cannot just set "Fuc(a1-3)GlcNAc(b1-" as the rejects, because "Gal(b1-4)GlcNAc(b1-" (the acceptor) is not the substructure of "Fuc(a1-3)GlcNAc(b1-" (the reject).

• rejects cannot be the same structure as acceptor.

If rejects is NULL, it means the enzyme does not reject any motifs.

product

The product is what the acceptor is transformed into by the enzyme. For glycosyltransferases (GTs), the product should have exactly one more residue than the acceptor. For glycoside hydrolases (GHs), the product should have exactly one less residue than the acceptor. If you have multiple possible products, put them in different rules. Take the same example above, if we want to add a Neu5Ac residue to the Gal residue of a ⁠Gal(b1-4)GlcNAc(b1-⁠ motif, we can set "Neu5Ac(a2-3)Gal(b1-4)GlcNAc(b1-" as the product.

For sulfotransferases, the product must have the same residue and linkage topology as the acceptor and add exactly one sulfate substituent.

requires

requires provides optional context motifs. Each entry contains motif and alignment; requirements are OR-valued, so at least one must match. An empty list means no additional context is required. Requirement motifs use sulfate-subset semantics, so an unsulfated requirement also matches the same topology with additional sulfates. Acceptor and reject motifs remain exact with respect to substituents during enzyme application.

See Also

enzyme()

Examples

library(glyrepr)

# Create a custom enzyme that adds a Neu5Ac residue to the Gal residue
make_enzyme(
  name = "MySiaT",
  rules = list(
    list(
      acceptor = "Gal(b1-4)GlcNAc(b1-",
      acceptor_alignment = "terminal",
      rejects = NULL,
      product = "Neu5Ac(a2-3)Gal(b1-4)GlcNAc(b1-"
    )
  ),
  type = "GT",
  species = "human"
)

# Create a custom enzyme that removes a Gal residue from a Gal(b1-4)GlcNAc(b1-` motif
make_enzyme(
  name = "MyGalH",
  rules = list(
    list(
      acceptor = "Gal(b1-4)GlcNAc(b1-",
      acceptor_alignment = "terminal",
      rejects = NULL,
      product = "GlcNAc(b1-"
    )
  ),
  type = "GH",
  species = "human"
)

# Create a custom enzyme with rejects
make_enzyme(
  name = "MySiaT",
  rules = list(
    list(
      acceptor = "Gal(b1-4)GlcNAc(b1-",
      acceptor_alignment = "terminal",
      rejects = c(
        # Reject a1-3 Fuc on GlcNAc
        "Fuc(a1-3)[Gal(b1-4)]GlcNAc(b1-",
        # Reject a1-2 Fuc on Gal
        "Fuc(a1-2)Gal(b1-4)GlcNAc(b1-"
      ),
      product = "Neu5Ac(a2-3)Gal(b1-4)GlcNAc(b1-"
    )
  ),
  type = "GT",
  species = "human"
)

# Create a custom enzyme with more than one rule
make_enzyme(
  name = "MySiaT",
  rules = list(
    list(
      acceptor = "Gal(b1-4)GlcNAc(b1-",
      acceptor_alignment = "terminal",
      rejects = NULL,
      product = "Neu5Ac(a2-3)Gal(b1-4)GlcNAc(b1-"
    ),
    # same acceptor, different product
    list(
      acceptor = "Gal(b1-4)GlcNAc(b1-",
      acceptor_alignment = "terminal",
      rejects = NULL,
      product = "Neu5Gc(a2-3)Gal(b1-4)GlcNAc(b1-"
    )
  ),
  type = "GT",
  species = "human"
)

Description

This function finds residues in glycans that match the product motifs of a glycosyltransferase or sulfotransferase and returns their node indices.

Usage

match_enzyme(glycans, enzyme, method = c("motif", "path"))

Arguments

Value

A list of integer vectors with the same length as glycans. Each integer vector contains node indices for residues added or modified by enzyme in the corresponding glycan.

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

glycan <- glyrepr::as_glycan_structure("Neu5Ac(a2-3)Gal(b1-3)GlcNAc(b1-")
match_enzyme(glycan, "ST3GAL3")
match_enzyme(glycan, "ST3GAL3", method = "path")

Description

Normalize selected Gal and Neu5Ac distributions in topological N-glycans. For the supported cases, glycans with the same numbers of antennae, Gal, and Neu5Ac are mapped to a common representative topology.

Usage

normalize_n_glycan(glycans)

Arguments

Value

A glyrepr::glycan_structure() vector of normalized glycans.

Why does this function exist?

trace_biosynthesis() and related functions support topological glycans. However, N-glycan annotations can assign residues to specific antennae even when the available data do not resolve those positions. These assignments can introduce unsupported distinctions into a biosynthesis network.

For example, consider the two glycans below.

Glycan 1:

Neu5Ac - Gal - GlcNAc
                      \
Neu5Ac - Gal - GlcNAc - Man
                           \
                            Man - GlcNAc - GlcNAc -
                           /
         Gal - GlcNAc - Man

Glycan 2:

Neu5Ac - Gal - GlcNAc
                      \
         Gal - GlcNAc - Man
                           \
                            Man - GlcNAc - GlcNAc -
                           /
Neu5Ac - Gal - GlcNAc - Man

Both are tri-antennary glycans with two sialic acids. When mass spectrometry data do not resolve which antennae carry the sialic acids, either topology may be used to annotate the same unresolved structure. Nevertheless, tracing these annotations can produce different biosynthetic paths and introduce redundant branches when multiple glycans are traced together.

Glycan 3 below illustrates the problem. It has one sialic acid and could serve as a precursor to a disialylated glycan. Of the two topologies above, only Glycan 2 can be obtained from Glycan 3 by adding a single sialic acid. Tracing Glycan 1 and Glycan 3 together therefore requires separate paths rather than connecting them through a single sialylation step. When the sialic acid positions are unresolved, this distinction reflects the annotation rather than evidence for separate biosynthetic routes.

Glycan 3:

         Gal - GlcNAc
                      \
         Gal - GlcNAc - Man
                           \
                            Man - GlcNAc - GlcNAc -
                           /
Neu5Ac - Gal - GlcNAc - Man

This function addresses the problem through a normalization convention. It applies rules to reposition Gal and Neu5Ac. In this example, mapping Glycan 1 to Glycan 2 allows the monosialylated Glycan 3 to connect to the disialylated representative through a single sialylation step. This convention does not imply that distinct isomers are biologically equivalent.

Algorithm

The algorithm repositions only Gal and Neu5Ac residues. To explain the algorithm, we name the branches for tri- and tetra-antennary glycans.

Tri-antennary:

A: GlcNAc
          \
B: GlcNAc - Man
                \
                 Man - GlcNAc - GlcNAc -
                /
C: GlcNAc - Man

Tetra-antennary:

A: GlcNAc
          \
B: GlcNAc - Man
                \
                 Man - GlcNAc - GlcNAc -
                /
C: GlcNAc - Man
          /
D: GlcNAc

First, Gal residues are repositioned in the following cases:

• Three antennae, one Gal: positioned on C.

• Three antennae, two Gal: positioned on A and C.

• Four antennae, two Gal: positioned on A and C.

Next, Neu5Ac residues are repositioned in the following cases:

• Three antennae, two Gal, one Neu5Ac: positioned on the C Gal.

• Three antennae, three Gal, one Neu5Ac: positioned on the C Gal.

• Three antennae, three Gal, two Neu5Ac: positioned on the A and C Gal.

• Four antennae, three Gal, one Neu5Ac: with Gal on A, B, and C, Neu5Ac is positioned on the C Gal.

• Four antennae, three Gal, two Neu5Ac: with Gal on A, B, and C, Neu5Ac residues are positioned on the A and C Gal.

• Four antennae, four Gal, two Neu5Ac: positioned on the A and C Gal.

Other distributions are left unchanged. Branch labels describe topology, not linkage positions or drawing order. For three antennae, C is on the Man with one antenna; A and B are interchangeable. For four antennae, A/B and C/D are the two pairs. With three Gal, A/B is the pair with two Gal, and C is the occupied antenna of the other pair.

Only concrete, topological structures are accepted. Use glyrepr::remove_linkages() explicitly before calling this function on intact or partial structures. Missing values are preserved.

Only simple GlcNAc, Gal-GlcNAc, and Neu5Ac-Gal-GlcNAc antennae are normalized. Non-N-glycans, other antenna counts, and structures with extended or modified antennae, other sialic acids, or floating parts or substituents are returned unchanged. Core fucose on the reducing-end GlcNAc and a bisecting GlcNAc on the central Man are preserved.

Examples

x <- glyparse::auto_parse(paste0(
  "Gal(b1-4)GlcNAc(b1-2)[GlcNAc(b1-6)]Man(a1-6)",
  "[GlcNAc(b1-2)Man(a1-3)]Man(b1-4)GlcNAc(b1-4)GlcNAc(b1-"
))
normalize_n_glycan(glyrepr::remove_linkages(x))

Description

Find biosynthetic paths from one glycan structure to another. The default method uses known enzyme rules in a forward breadth-first search. To infer structure-driven paths without enzyme specificity, use path_biosynthesis_virtual().

Usage

path_biosynthesis(
  from,
  to,
  enzymes = NULL,
  max_steps = 10,
  filter = NULL,
  max_virtual_steps = 0L
)

Arguments

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 the 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.

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-"

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.

Examples

library(glyrepr)
library(glyparse)

# Find shortest path
from <- "Gal(b1-4)GlcNAc(b1-"
to <- "Neu5Ac(a2-6)Gal(b1-4)GlcNAc(b1-"
path <- path_biosynthesis(from, to, enzymes = "ST6GAL1", max_steps = 3)

# View the path
igraph::as_data_frame(path, what = "edges")

Description

Infer every structure-driven biosynthetic path from from to to by trimming terminal residues and sulfate groups from to backward to from. Unlike path_biosynthesis(), this does not require known enzyme rules.

Usage

path_biosynthesis_virtual(from, to, enzymes = NULL, annotate_enzymes = FALSE)

Arguments

Value

A glyenzy_virtual_biosynthesis_network object inheriting from glyenzy_biosynthesis_network and igraph::igraph(). Vertices contain IUPAC-condensed strings in name and a logical target attribute indicating the target glycan. At most one directed edge connects each substrate and product. Edges have a forward step, is_virtual = TRUE, a virtual-enzyme enzyme label, and a list-valued enzymes attribute. When annotate_enzymes is TRUE, enzymes contains every candidate concrete enzyme for each transition; otherwise each element is empty.

Virtual enzymes

Each edge is named for the residue added by that step. Intact glycans include the linkage anomer and acceptor position, so a beta-1,4-linked GlcNAc is labeled "b4GlcNAcT". Topological glycans omit linkage information and use "GlcNAcT". Generic topological glycans use their preserved generic residue names, such as "HexNAcT".

Sulfation is represented as its own atomic transition. Sulfate additions at positions 3 and 6 use "3SulfoT" and "6SulfoT"; an unknown or other position uses "?SulfoT". A sulfated terminal residue is therefore desulfated before the residue itself can be trimmed.

Virtual tracing starts N-glycans at the N-glycan core and all other glycans at their reducing-end root residue. Sulfates are removed from these automatically selected starts. In path_biosynthesis_virtual(), the explicit from glycan is always the virtual starting structure, including any sulfate groups it contains; those sulfates must also occur in to. These networks represent structural possibilities rather than biological feasibility.

Generic structures do not retain glycan-class metadata. A generic structure matching the N-glycan-core topology is therefore assumed to be an N-glycan; use path_biosynthesis_virtual() with an explicit from when that topology belongs to another glycan class.

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.

Examples

virtual_path <- path_biosynthesis_virtual(
  "Gal(b1-3)GalNAc(a1-",
  "GlcNAc(b1-4)Gal(b1-3)GalNAc(a1-"
)

annotated_path <- path_biosynthesis_virtual(
  "Gal(b1-3)GalNAc(a1-",
  "GlcNAc(b1-4)Gal(b1-3)GalNAc(a1-",
  annotate_enzymes = TRUE
)

Description

Print method for glyenzy_enzyme objects

Usage

## S3 method for class 'glyenzy_enzyme'
print(x, ...)

Arguments

Description

Calculate the ratio between the motif quantification of enzyme products and substrates for glycosyltransferases and sulfotransferases. For glycoproteomics data, ratios are calculated independently for each glycosite.

Usage

product_substrate_ratio(exp, enzymes)

Arguments

Details

Each rule contributes its product and acceptor motif once. Quantifications are summed across all rules of an enzyme before division, including when multiple rules contain the same motif. Rule rejects and requires are ignored. Ratios with zero substrate quantification are returned as NA.

Motifs are matched strictly for intact glycan structures and leniently for partial or reduced structures, consistent with other glyenzy motif-based functions.

Value

A plain SummarizedExperiment::SummarizedExperiment() with a product_substrate_ratio assay. For glycomics data, rowData() contains enzyme. For glycoproteomics data, it contains enzyme, protein, and protein_site. colData() and metadata are preserved from exp.

References

Bao B, Kellman BP, Chiang AWT, et al. (2021). Correcting for sparsity and interdependence in glycomics by accounting for glycan biosynthesis. Nature Communications, 12, 4988. doi:10.1038/s41467-021-25183-5

Examples

exp <- glyexp::real_experiment2[seq_len(10), seq_len(3)]
product_substrate_ratio(exp, "ST6GAL1")

Description

Reconstruct 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

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")

Description

Reconstruct every structure-driven biosynthetic path for one or more glycans by trimming terminal residues and sulfate groups backward. Unlike trace_biosynthesis(), this does not require known enzyme rules.

Usage

trace_biosynthesis_virtual(glycans, enzymes = NULL, annotate_enzymes = FALSE)

Arguments

Value

A glyenzy_virtual_biosynthesis_network object inheriting from glyenzy_biosynthesis_network and igraph::igraph(). Vertices contain IUPAC-condensed strings in name and a logical target attribute indicating whether each vertex is a target glycan. At most one directed edge connects each substrate and product. Edges have a forward step, is_virtual = TRUE, a virtual-enzyme enzyme label, and a list-valued enzymes attribute. When annotate_enzymes is TRUE, enzymes contains every candidate concrete enzyme for each transition; otherwise each element is empty.

Virtual enzymes

Each edge is named for the residue added by that step. Intact glycans include the linkage anomer and acceptor position, so a beta-1,4-linked GlcNAc is labeled "b4GlcNAcT". Topological glycans omit linkage information and use "GlcNAcT". Generic topological glycans use their preserved generic residue names, such as "HexNAcT".

Sulfation is represented as its own atomic transition. Sulfate additions at positions 3 and 6 use "3SulfoT" and "6SulfoT"; an unknown or other position uses "?SulfoT". A sulfated terminal residue is therefore desulfated before the residue itself can be trimmed.

Virtual tracing starts N-glycans at the N-glycan core and all other glycans at their reducing-end root residue. Sulfates are removed from these automatically selected starts. In path_biosynthesis_virtual(), the explicit from glycan is always the virtual starting structure, including any sulfate groups it contains; those sulfates must also occur in to. These networks represent structural possibilities rather than biological feasibility.

Generic structures do not retain glycan-class metadata. A generic structure matching the N-glycan-core topology is therefore assumed to be an N-glycan; use path_biosynthesis_virtual() with an explicit from when that topology belongs to another glycan class.

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-"

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.

Examples

library(glyrepr)
library(glyparse)

virtual_path <- trace_biosynthesis_virtual(
  "GlcNAc(b1-4)Gal(b1-3)GalNAc(a1-"
)

annotated_path <- trace_biosynthesis_virtual(
  "GlcNAc(b1-4)Gal(b1-3)GalNAc(a1-",
  annotate_enzymes = TRUE
)

Description

Visualize where an enzyme adds or modifies residues in a glycan structure.

Usage

view_enzyme(glycan, enzyme)

Arguments

Details

view_enzyme() matches one enzyme against one glycan with the same matching rules used by match_enzyme(), then draws the glycan with the matched residues highlighted.

Value

A ggplot object returned by glydraw::draw_cartoon(). If no match is found, the glycan is drawn without highlighted residues and a cli alert is emitted.

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-"

See Also

match_enzyme(), glydraw::draw_cartoon()

Examples

glycan <- glyrepr::as_glycan_structure("Neu5Ac(a2-3)Gal(b1-3)GlcNAc(b1-")

## Not run: 
view_enzyme(glycan, "ST3GAL3")

## End(Not run)