01
Transcription and splicing
Transcript production, exon architecture and exon-junction-complex context.
11 — Translation small-world engine
Translation is a network,
not a single reaction.
A suppressor tRNA enters a connected system of decoding, termination, surveillance, ribosome traffic, stress signalling and protein folding. KritRNA’s small-world engine is intended to model those interactions before a candidate is advanced.
Biological layers
From transcript birth to protein fate
The model is organised as connected layers rather than a flat feature list. A perturbation at the stop codon can propagate upstream and downstream through the translation system.
02
Export and initiation
Message availability, initiation competence and ribosome loading.
03
Elongation
Codon demand, endogenous tRNA supply, decoding velocity and local pausing.
04
Collision and stress
Ribosome queues, collision sensing and activation thresholds for stress pathways.
05
Termination
Competition among suppressor tRNA, eRF1/eRF3 and local stop-context effects.
06
Recycling
Post-termination ribosome recycling and availability for subsequent translation rounds.
07
NMD and quality control
UPF-dependent surveillance, ribosome rescue and no-go/quality-control responses.
08
Folding and proteostasis
Consequences of altered translation timing for protein folding, chaperones and degradation.
Molecular network
The candidate is only one node
Each molecular player creates a possible bottleneck, source of competition or safety signal. The network model is intended to make those dependencies explicit.
Ribosome
The central translation machine and the site of the candidate intervention.
Suppressor tRNA
The designed perturbation introduced into the network.
Aminoacyl-tRNA synthetase
Charges the tRNA and determines whether amino-acid identity is preserved.
eRF1 / eRF3
Compete for recognition and termination at stop codons.
ABCE1
Supports recycling after termination.
UPF1 / UPF2 / UPF3
Core nonsense-mediated-decay machinery that influences transcript persistence.
PELO / HBS1
Ribosome-rescue machinery relevant to stalled or problematic translation events.
ZAKα / GCN2 / PERK / PKR / HRI
Stress sensors connecting ribosome or cellular perturbation to the integrated stress response.
eIF2α / ATF4
Downstream stress-response nodes affecting global translation and adaptation.
Chaperones and proteostasis systems
Determine whether a restored polypeptide reaches a stable and useful state.
Computational backbone
Different biological questions need different mathematics
No single model class can credibly represent sequence design, ribosome traffic, network diffusion and stress dynamics. The architecture combines methods according to the mechanism being modelled.
Small-world graph metrics
Represent high local clustering with short paths between translation, surveillance and stress modules.
Graph Laplacian diffusion
Propagate a candidate perturbation across connected biological modules.
Markov transitions
Model probabilistic progression among decoding, pausing, termination, readthrough and rescue states.
Queueing models
Represent ribosome traffic, spacing, queues and collision probability on an mRNA.
ODE thresholds
Describe stress-response activation and recovery as coupled dynamic systems.
Stochastic simulation
Capture cell-to-cell and event-to-event variability that deterministic averages can hide.
These are architecture components under development. The public website does not claim that every module is fully trained, calibrated or experimentally validated.
Data architecture
Public knowledge becomes useful only after biological alignment
Databases, structures, expression records and experiments describe different parts of the system. The platform is intended to connect them through explicit biological entities and provenance.
01
tRNA sequence and annotation
GtRNAdb, tRNAscan-SE outputs and curated species-specific records.
02
tRNA modification biology
MODOMICS and primary literature on identity, processing and modification.
03
Transcript architecture
Ensembl, APPRIS and exon-junction / isoform context.
04
Termination context
Stop-codon neighbourhoods, +4 base effects and curated readthrough literature.
05
Expression and abundance
Expression Atlas and context-specific RNA/tRNA abundance datasets.
06
Ribosome behaviour
Ribo-seq and other translation-profiling datasets where assay quality supports use.
07
Structure and thermodynamics
RNA secondary/tertiary prediction, resolved structures and simulation-derived features.
08
Experimental outcomes
Literature-curated assays and future KritRNA wet-lab results stored with provenance.
Dataset philosophy
Schema before scale
A dual-luciferase readthrough value, Western-blot densitometry and functional protein rescue may describe related biology, but they are not the same label. Pooling them naively produces noise that looks like signal.
KritRNA’s dataset philosophy is to preserve assay context, separate outcome types and record uncertainty before increasing dataset size.
Model the translation system, then challenge it at the bench.
The small-world engine is designed to generate testable system-level hypotheses. Experimental evidence remains the authority.