Experiment Types • glyexp

Welcome to the world of glycoverse experiments! This vignette will walk you through the four distinct experiment types in the glyexp package: “glycomics”, “glycoproteomics”, “traitomics”, and “traitproteomics”. Think of these as different flavors of glycobiology data, each with their own personality and purpose. You can check what type your experiment is using get_exp_type().

library(glyexp)
library(glyrepr)

What’s the big deal with experiment types?

When we talk about experiment types, we’re really talking about two things:

A label stored in the exp_type field of your experiment() object’s metadata
The expected structure of your variable information table

Here’s the thing: when your experiment has a specific type, all the functions in glycoverse know what to expect from your data structure. It’s like having a contract between your data and the analysis functions. Don’t worry though— these are more like friendly guidelines than iron-clad rules. We’ll show you what this means in practice with some examples.

Quick tip: If you’re not diving into glydet or glymotif, you can safely skip the traitomics and traitproteomics sections and focus on the first two types.

Meet the four experiment types

The glyexp package recognizes four distinct experiment types:

“glycomics” - Your classic glycan-focused experiments
“glycoproteomics” - When glycans meet proteins
“traitomics” - Derived traits from glycan data
“traitproteomics” - Derived traits at the protein level

The first two are your bread-and-butter glycobiology experiments. When you use glyread to import your data, it’ll automatically figure out whether you’re dealing with glycomics or glycoproteomics and set the type accordingly— pretty neat, right?

The last two are special creatures unique to the glycoverse ecosystem. They’re what you get when functions from glymotif or glydet work their magic on your data. Let’s dive into each type to see what makes them tick.

Glycomics

Let’s start with the most straightforward type: “glycomics”. This is your classic glycan-focused experiment where each row in the variable information table represents a single glycan (or a spectrum match if you’re working with MS data).

What to expect in your variable information table:

glycan_composition: The glycan’s composition as a glyrepr::glycan_composition() object
glycan_structure: (Optional) The glycan’s structure as a glyrepr::glycan_structure() object

Here’s what this looks like in practice:

get_var_info(real_experiment2)
#> # A tibble: 67 × 3
#>    variable glycan_composition                  
#>    <chr>    <comp>                              
#>  1 V1       Man(3)GlcNAc(3)                     
#>  2 V2       Man(3)GlcNAc(7)                     
#>  3 V3       Man(5)GlcNAc(2)                     
#>  4 V4       Man(4)Gal(2)GlcNAc(4)Neu5Ac(2)      
#>  5 V5       Man(3)Gal(1)GlcNAc(3)               
#>  6 V6       Man(3)Gal(2)GlcNAc(4)Fuc(2)         
#>  7 V7       Man(3)GlcNAc(3)Fuc(1)               
#>  8 V8       Man(3)GlcNAc(4)                     
#>  9 V9       Man(3)Gal(2)GlcNAc(5)Neu5Ac(1)      
#> 10 V10      Man(3)Gal(1)GlcNAc(5)Fuc(1)Neu5Ac(1)
#> # ℹ 57 more rows
#> # ℹ 1 more variable: glycan_structure <struct>

Glycoproteomics

Now we step up to “glycoproteomics”—where things get more interesting! Here, each row represents a glycopeptide or glycoform (or their peptide spectrum matches). Think of it as glycomics data with protein context, telling you not just what glycans are there, but where they’re attached.

Your variable information table should include:

protein: The protein UniProt accession (character string)
protein_site: The glycosylation site position on the protein (integer)
glycan_composition: The glycan composition as a glyrepr::glycan_composition() object
glycan_structure: (Optional) The glycan structure as a glyrepr::glycan_structure() object

Here’s a real example to illustrate:

get_var_info(real_experiment)
#> # A tibble: 4,262 × 8
#>    variable peptide   peptide_site protein protein_site gene  glycan_composition
#>    <chr>    <chr>            <int> <chr>          <int> <chr> <comp>            
#>  1 GP1      JKTQGK               1 P08185           176 SERP… Hex(5)HexNAc(4)Ne…
#>  2 GP2      HSHNJJSS…            5 P04196           344 HRG   Hex(5)HexNAc(4)Ne…
#>  3 GP3      HSHNJJSS…            5 P04196           344 HRG   Hex(5)HexNAc(4)   
#>  4 GP4      HSHNJJSS…            5 P04196           344 HRG   Hex(5)HexNAc(4)Ne…
#>  5 GP5      HJSTGCLR             2 P10909           291 CLU   Hex(6)HexNAc(5)   
#>  6 GP6      HSHNJJSS…            5 P04196           344 HRG   Hex(5)HexNAc(4)Ne…
#>  7 GP7      HSHNJJSS…            6 P04196           345 HRG   Hex(5)HexNAc(4)   
#>  8 GP8      HSHNJJSS…            5 P04196           344 HRG   Hex(5)HexNAc(4)dH…
#>  9 GP9      HSHNJJSS…            5 P04196           344 HRG   Hex(4)HexNAc(3)   
#> 10 GP10     HSHNJJSS…            5 P04196           344 HRG   Hex(4)HexNAc(4)Ne…
#> # ℹ 4,252 more rows
#> # ℹ 1 more variable: glycan_structure <struct>

Traitomics

Here’s where things get uniquely glycoverse! Both traitomics and traitproteomics are special experiment types that emerge when you transform your data using glydet::derive_traits() or glymotif::quantify_motifs().

Starting with traitomics: when you feed a glycomics experiment to derive_traits(), you get back a traitomics experiment. Instead of measuring individual glycans in each sample, you’re now looking at derived traits—think of it as a higher-level summary of your glycan data.

The variable information table is pretty flexible here—no mandatory columns. However, you’ll typically see:

A trait column (if using derive_traits())
A motif column (if using quantify_motifs())

Traitproteomics

Last but not least, we have traitproteomics—the most sophisticated of our experiment types. When you apply derive_traits() to a glycoproteomics experiment, you get a traitproteomics experiment back.

Here’s the key insight: instead of tracking individual glycans at each protein site, you’re now measuring derived traits at each site. Think of each glycosylation site as having its own little glycome, and you’re summarizing the characteristics of that mini-glycome.

Essential columns in the variable information table:

protein: The protein UniProt accession (character string)
protein_site: The glycosylation site position (integer)

Plus you’ll get:

A trait column (from derive_traits())
A motif column (from quantify_motifs())

Here’s a key difference: in glycoproteomics, different sites can have completely different glycan repertoires. But in traitproteomics, every site gets scored on the same set of traits or motifs—it’s like having standardized metrics across all your glycosylation sites.

Do you need to worry about experiment types?

The short answer? Not really! Experiment types work behind the scenes quite intuitively. You’ll mostly encounter them in two situations:

Reading documentation - When you see references to specific experiment types
Error messages - Like “Expecting a glycoproteomics experiment, but got a traitproteomics experiment”

The reason these types matter is that different functions have different expectations. For instance:

glymotif::quantify_motifs() wants glycomics or glycoproteomics data
glystats::gly_enrich_go() expects glycoproteomics or traitproteomics experiments

When you might need to care: The main scenario is when you’re manually creating an experiment object using experiment(). In that case, you’ll want to set the exp_type field to match your data structure.

Important warning: Once you’ve created an experiment object, resist the temptation to manually fiddle with the exp_type field. This will break the implicit contract between your data and the glycoverse functions—and nobody wants that!