Does letting each node choose its own depth fix GNN over-smoothing?
Letting each drug or protein node decide how far to gather information from its neighbours kept graph features distinct and slightly improved drug-target prediction.
Source
iGRLDTI: an improved graph representation learning method for predicting drug-target interactions over heterogeneous biological information network
Study at a glance
- Design
- Computational / modelling — Graph of drugs and protein targets with learned node features; node-dependent smoothing depth, then a gradient-boosted tree classifier; 10-fold cross-validation against five published methods.
- N
- N=1923 · 1923 known drug-target interactions among 549 drugs and 424 targets, plus an equal number of randomly paired negatives (a 1:10 imbalanced setting was also run).
- Population
- Drugs and protein targets from DrugBank 3.0 and HPRD
- Outcome
- AUC, AUPR and F1 for predicting drug-target interactions; mean absolute distance between node embeddings as an over-smoothing measure
Structured fields used in claim comparison tables when every cited study has a complete layer.
What they did
The authors built a network linking 549 drugs and 424 protein targets through 1923 known interactions, giving each drug a feature vector from its chemical structure and each protein one from its sequence. Instead of using the same number of message-passing steps for every node, their method iGRLDTI picks a node-specific depth (node-dependent local smoothing) and averages features up to that depth; a gradient-boosted decision tree then classifies drug-target pairs. They compared it with five published methods under 10-fold cross-validation and ran ablations.
What they found
iGRLDTI had the best AUC and AUPR, but margins over the strongest competitors were small (about one to four percentage points on balanced data). In a harder 1:10 imbalanced setting its AUPR advantage was larger, up to 25.5% over one baseline. Ablations showed that using the graph at all helped a lot over raw features, and the adaptive depth added a further gain of a few percent over a standard GNN; its embeddings were also measurably less similar to each other, meaning less over-smoothing.
The limits
What it doesn't show
The benchmark is small and old, and negative examples were random unlabelled pairs, some of which may be true but undiscovered interactions. Most gains over the best baselines are only around one to two points and come from a single dataset, so generality is untested. The case-study validations use later DrugBank versions, not new lab experiments, and the authors note that nodes lacking biological features, such as newly discovered drugs, will be predicted poorly.
Key terms
- Over-smoothing
- When many rounds of graph message passing make all node representations look alike, so the model can no longer tell nodes apart.
- Propagation depth
- How many hops away a node gathers information from in a graph neural network.
- Node-dependent local smoothing
- A rule that picks a separate propagation depth for each node, stopping before its features converge to the graph-wide average.
- Heterogeneous network
- A graph containing more than one kind of node, here drugs and protein targets.
- Gradient boosting decision tree
- A classifier that builds many small decision trees in sequence, each correcting the errors of the previous ones.
Flashcards
0 of 9 answers reviewed
Research intelligence for this paper
See its role on concept claims, tensions it is part of, placement history, and related discoveries.
Quiz yourself
What does over-smoothing in a graph neural network refer to?
Common questions
Why not just use a deeper graph network?
Deeper propagation mixes in so much neighbourhood information that node features become nearly identical; the paper's smoothness metric falls as depth grows.
Are the negatives in this benchmark truly non-interacting?
No. They are random drug-target pairs with no recorded interaction, so some may be real interactions that simply have not been found yet.
How large is the improvement?
Modest on balanced data, typically a point or two over the best baselines, and larger in AUPR when positives are rare.
More on Graph neural networks