Lysosomal Storage Diseases Project

Overview

This project organizes curation of lysosomal storage diseases (LSDs) in dismech, with a specific focus on where to lump, where to split, and where to use a project umbrella rather than forcing everything into one root disease file.

Lysosomal storage disease is a useful project label, but it is usually too broad to be a single good dismech disease entry. The disorders share a common cell-biologic theme — failure of lysosomal degradation, trafficking, or substrate clearance — but they often diverge sharply in:

• the stored substrate
• the defective enzyme / transporter / accessory factor
• the dominant cell types and organs affected
• treatment logic
• age-of-onset structure and subtype conventions

So the right default in dismech is:

• project umbrella at the LSD level
• disease-level files for biologically distinct roots
• subtypes within a root when the proximal mechanism is still shared

This document is a project brief, not a claim that all lysosomal diseases should collapse into one disease YAML.

Existing LSD-ish KB Anchors

Current kb/disorders/ files already touching the lysosomal-storage space:

Clearly in scope:

• Fabry_Disease.yaml
• Gaucher_Disease.yaml
• Pompe_Disease.yaml
• Niemann_Pick_Disease_Type_C.yaml
• Krabbe_Disease.yaml
• Metachromatic_Leukodystrophy.yaml
• Tay-Sachs_Disease.yaml
• Beta_Mannosidosis.yaml
• Salla_Disease.yaml
• Hurler_syndrome.yaml
• Mucopolysaccharidosis.yaml
• Neuronal_Ceroid_Lipofuscinosis.yaml

Borderline / adjacent but still useful to track here:

• Danon_disease.yaml
• lysosome/autophagic vacuolar biology is relevant, but this is not the same curation bucket as classic substrate-storage disorders.

Important current modeling signal from the repo:

• Hurler_syndrome.yaml already exists as a split disease root.
• Mucopolysaccharidosis.yaml also exists as a broader umbrella/family.
• Neuronal_Ceroid_Lipofuscinosis.yaml exists as an umbrella root.
• Sanfilippo is currently mentioned under Mucopolysaccharidosis, not yet as its own standalone disease file in kb/disorders/.

That means the repo already wants a mixed strategy: broad family/project anchors plus selected biologically strong disease roots.

Recommended Modeling Strategy

Recommendation: project umbrella + family roots + selective subtype splits

Use three levels deliberately:

1. Project level — projects/LYSOSOMAL_STORAGE_DISEASES.md - tracks the whole class - records split/lump rules - organizes priorities

2. Disease-family or disease-root files in kb/disorders/ - used when a disease has a stable mechanistic trunk and established clinical identity

3. Subtype structure - used when named subtypes mostly share the same proximal defect and differ more by severity, age, tissue weighting, or branch-point consequences than by a truly different mechanistic trunk

Working Lump / Split Rules

Split into separate top-level disease files when:

• the causal gene / enzyme / transporter / accessory factor changes and that change materially alters the proximal mechanism
• the stored substrate class changes in a way that changes the mechanistic trunk (for example sphingolipid vs glycogen vs glycosaminoglycan vs free sialic acid)
• the organelle/process framing changes enough that one file would blur the biology (for example cholesterol trafficking in NPC vs acid sphingomyelinase deficiency in Niemann-Pick A/B)
• the treatment logic materially differs (enzyme replacement, substrate reduction, chaperone therapy, HSCT, CNS-targeted gene therapy, etc.)
• the literature and clinical practice treat the entity as a distinct disease family, not just a subtype label

Keep within one disease file and use internal subtype structure when:

• the proximal defect is the same, and subtypes mainly reflect severity, age of onset, neuroinvolvement, or tempo
• the named subtype labels are clinically useful but do not imply a new mechanistic trunk
• splitting would mostly produce ontology noise rather than more accurate pathographs

Use a family umbrella plus separate roots when:

• the family has a shared biochemical identity but several clearly distinct enzyme-level disease roots
• the family is useful for project organization but not as a single final disease pathograph

Family-Specific Guidance

1. Mucopolysaccharidoses (MPS)

Do not force all MPS into one disease file.

Recommended:

• keep Mucopolysaccharidosis.yaml as a family umbrella / project anchor
• maintain separate root files for major enzyme-defined diseases

Good split candidates / existing anchors:

• Hurler_syndrome.yaml (already split)
• Hunter syndrome
• Sanfilippo syndrome
• Morquio syndrome
• Maroteaux-Lamy syndrome
• Sly syndrome

Within-family rule:

• split at the major enzyme/disease level
• but do not automatically split every subtype letter into a new file unless the subtype has a meaningfully distinct mechanistic trunk

Specific recommendation:

• treat Sanfilippo syndrome as a strong candidate for a root file
• treat Sanfilippo A/B/C/D as likely subtypes under that root first, not four premature top-level files by default

2. Neuronal ceroid lipofuscinoses (NCL / Batten spectrum)

Recommended:

• keep Neuronal_Ceroid_Lipofuscinosis.yaml as the umbrella root
• do not start by creating standalone root files for every age label

Key rule:

• labels like infantile / late-infantile / juvenile / adult NCL are often best treated as subtypes, not automatic separate roots

Specific recommendation:

• adult neuronal ceroid lipofuscinosis should default to subtype treatment under NCL rather than its own root entry unless the literature supports a clearly distinct mechanistic trunk that would justify separation

3. Gaucher disease

Recommended:

• keep one Gaucher root file
• represent type 1 / 2 / 3 as subtypes or internal branches, not separate top-level files by default

Why:

• the proximal GBA / glucocerebroside-storage mechanism is shared
• the main differences are neurologic involvement and tempo, not a totally different mechanistic trunk

4. Niemann-Pick family

Recommended:

• keep NPC as a separate root (Niemann_Pick_Disease_Type_C.yaml already exists)
• do not lump NPC together with acid sphingomyelinase deficiency forms just because the family name overlaps

Why:

• NPC is a cholesterol-trafficking / lysosomal export problem
• Niemann-Pick A/B are acid sphingomyelinase deficiency disorders with a different proximal defect

So:

• split NPC from Niemann-Pick A/B
• family label can remain project-level only

5. GM2 gangliosidosis branch

Recommended:

• do not make a generic GM2 root the default endpoint unless needed for project organization
• prefer separate root files for:
• Tay-Sachs disease
• Sandhoff disease

Why:

• HEXA vs HEXB matters mechanistically and clinically
• the shared ganglioside theme is useful, but gene/enzyme-specific roots are clearer in dismech

6. Leukodystrophy / sphingolipid storage edge cases

Recommended:

• keep Krabbe_Disease.yaml and Metachromatic_Leukodystrophy.yaml as separate roots
• do not lump them just because both are lysosomal leukodystrophies

Why:

• galactocerebrosidase deficiency vs arylsulfatase A / sulfatide handling are different trunks

7. Borderline lysosomal / autophagic disorders

Danon_disease.yaml is worth tracking in this project, but it should be treated as an edge case rather than the template for classic LSD curation.

Why:

• the lysosome/autophagy axis is central
• but the disease family framing, tissue weighting, and treatment context differ from many classic substrate-storage disorders

Current Modeling Stance

Strong root files already present or clearly justified

• Fabry disease
• Gaucher disease
• Pompe disease
• Niemann-Pick disease type C
• Krabbe disease
• Metachromatic leukodystrophy
• Tay-Sachs disease
• Beta mannosidosis
• Salla disease
• Hurler syndrome
• Neuronal ceroid lipofuscinosis
• Mucopolysaccharidosis (as umbrella/family anchor, not final lump)

Strong next-wave disease roots

• Hunter syndrome
• Sanfilippo syndrome
• Morquio syndrome
• Sandhoff disease
• Niemann-Pick disease type A/B branch
• Wolman disease
• Farber disease
• Alpha-mannosidosis
• GM1 gangliosidosis

Likely subtype-first rather than root-first

• adult neuronal ceroid lipofuscinosis
• Sanfilippo A/B/C/D when a Sanfilippo root does not yet exist
• Gaucher type 1 / 2 / 3
• late-infantile / juvenile / adult NCL labels in isolation

Practical Curation Rules for This Project

• Do not let a family label stand in for a good mechanistic disease root.
• Do not create a new root file just because medicine has a subtype name.
• Prefer enzyme/gene/substrate-defined roots over purely age-labeled roots.
• Keep phenotypes downstream from the storage / trafficking defect rather than turning severity labels into fake mechanisms.
• When the same trunk feeds several subtype branches, use one file + subtype structure instead of multiple thin YAMLs.
• When distinct biochemical lesions produce distinct treatment and mechanism logic, split.

Initial Execution Plan

1. Keep this project file as the organizing umbrella.
2. Audit current lysosomal-related entries for consistency of lump/split style.
3. Normalize the MPS strategy: - retain Mucopolysaccharidosis.yaml as umbrella - use disease-level roots for major enzyme-defined forms - use subtype structure under those roots where appropriate
4. Normalize the NCL strategy: - umbrella root first - age-form labels as subtype-first unless clearly justified otherwise
5. Add missing high-value roots in this order: - Hunter syndrome - Sanfilippo syndrome - Sandhoff disease - Morquio syndrome - Wolman disease
6. Reassess whether a reusable lysosomal mechanism module is worth creating only after 3-5 more roots exist.

Open Questions

• Should Mucopolysaccharidosis.yaml stay a lightweight umbrella, or should it become more explicitly a family-summary page with has_subtypes pointing to major enzyme-defined roots?
• Should Sanfilippo be rooted first as one disease family with subtype letters, or should A/B/C/D split immediately once a Sanfilippo root exists?
• Should Neuronal_Ceroid_Lipofuscinosis.yaml remain the main umbrella while CLN gene-specific branches stay internal, or is there enough repeated divergence to justify selected separate roots later?
• How much lysosome/autophagy edge territory belongs here versus adjacent disease-family projects?

STATUS

Existing Anchors

• [x] Identified current lysosomal-related disorder files already present in the KB
• [x] Recorded current umbrella/root tension in MPS and NCL
• [x] Wrote initial lump/split strategy for LSD curation

Next Curation Targets

• [ ] Hunter syndrome
• [ ] Sanfilippo syndrome
• [ ] Sandhoff disease
• [ ] Morquio syndrome
• [ ] Wolman disease
• [ ] Farber disease
• [ ] Alpha-mannosidosis

NOTES

2026-04-14

• Started LYSOSOMAL_STORAGE_DISEASES project brief.
• Chose a project umbrella + disease-root + subtype strategy rather than a single giant LSD disease file.
• Recorded that current repo state already implies mixed granularity: Hurler_syndrome.yaml is split, while Mucopolysaccharidosis.yaml and Neuronal_Ceroid_Lipofuscinosis.yaml act as broader umbrellas.
• Explicitly treated adult neuronal ceroid lipofuscinosis as subtype-first, not an automatic standalone root.