Jia et al.

BMC Bioinformatics (2024) 25:214 BMC Bioinformatics

[Link]

RESEARCH Open Access

A deep learning framework for predicting

disease‑gene associations with functional
modules and graph augmentation
Xianghu Jia1†, Weiwen Luo1†, Jiaqi Li1†, Jieqi Xing1, Hongjie Sun1, Shunyao Wu1* and Xiaoquan Su1*

†
Xianghu Jia, Weiwen Luo and
Jiaqi Li have contributed equally,
Abstract
co-first authors. Background: The exploration of gene-disease associations is crucial for understand-
*Correspondence: ing the mechanisms underlying disease onset and progression, with significant impli-
wushunyao@[Link]; cations for prevention and treatment strategies. Advances in high-throughput biotech-
suxq@[Link]
nology have generated a wealth of data linking diseases to specific genes. While graph
1
College of Computer Science representation learning has recently introduced groundbreaking approaches for pre-
and Technology, Qingdao
University, Qingdao 266071, dicting novel associations, existing studies always overlooked the cumulative impact
Shandong, China of functional modules such as protein complexes and the incompletion of some
important data such as protein interactions, which limits the detection performance.
Results: Addressing these limitations, here we introduce a deep learning framework
called ModulePred for predicting disease-gene associations. ModulePred performs
graph augmentation on the protein interaction network using L3 link prediction
algorithms. It builds a heterogeneous module network by integrating disease-gene
associations, protein complexes and augmented protein interactions, and develops
a novel graph embedding for the heterogeneous module network. Subsequently,
a graph neural network is constructed to learn node representations by collectively
aggregating information from topological structure, and gene prioritization is carried
out by the disease and gene embeddings obtained from the graph neural network.
Experimental results underscore the superiority of ModulePred, showcasing the effec-
tiveness of incorporating functional modules and graph augmentation in predicting
disease-gene associations. This research introduces innovative ideas and directions,
enhancing the understanding and prediction of gene-disease relationships.
Keywords: Gene-disease associations, Deep learning, Graph augmentation, Protein
complexes, Graph neural networks

Introduction
Gene mutations or genetic abnormalities play a pivotal role in the pathogenesis of vari-
ous diseases. Consequently, uncovering the associations between genes and diseases is
imperative to elucidate the underlying molecular mechanisms and enhance healthcare.
While linkage analysis and genome-wide association studies are capable of detecting
biomarkers, such as single nucleotide polymorphisms (SNPs), by examining genetic

© The Author(s) 2024. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits
use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original
author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third
party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the mate-
rial. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or
exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://
[Link]/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver ([Link]
main/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.
Jia et al. BMC Bioinformatics (2024) 25:214 Page 2 of 14

variations within human populations, these approaches are time and resource-intensive
due to the necessity of analyzing numerous false positives [1]. Moreover, these methods
primarily focus on direct connections between genotypes and phenotypes, thereby over-
looking the complex interactions between molecules [2].
Recent years, computational methods rooted in molecular networks have emerged as
a prominent approach to complement and enhance linkage analysis and genome-wide

Fig. 1 An overview of our proposed approach. Firstly, Data augmentation was performed on the protein–
protein interaction (PPI) network with L3 principle (A). Then, by integrating augmented PPI network, protein
complexes and disease-gene associations (B), a heterogeneous module network was built (C). Subsequently,
initial low-dimensional embeddings were obtained by graph representation (D) for the heterogeneous
module network and candidate genes were generated for each disease (E). Furthermore, a graph neural
network was constructed to learn better representations by collectively aggregating information from
topological structure (F). Finally, for each disease, the candidate genes were scored and re-ranked based on
the embeddings generated by the graph neural network (G)
Jia et al. BMC Bioinformatics (2024) 25:214 Page 3 of 14

association studies, providing valuable insights into disease gene prediction [3–5]. The
primary objective is to extract topological features that precisely capture the intricate
connections between genes and diseases, including measures of topological similarity
between genes and diseases [6–8], as well as other artificially extracted features [9–11].
Notably, graph embedding methods such as node2vec and graph neural networks like
graph convolutional network (GCN) have witnessed extensive application in gene-dis-
ease association mining, showcasing commendable performance by automatically dis-
covering potent latent features [12, 13]. Despite significant strides in existing research,
certain issues impede detection performance, including the oversight in investigating
cooperative relationships among molecules. For instance, in cellular activities, proteins
often depend on collaborative interactions within protein complexes to execute specific
functions [14]. Additionally, the effectiveness of disease gene prediction faces substantial
hindrance due to the incompleteness of existing molecular networks, notably the protein
interaction network, which lacks experimental validation for numerous interactions.
This paper introduces a novel paradigm centered on modules to encapsulate coop-
erative relationships among molecules, particularly focusing on protein complexes. We
present ModulePred, an advanced deep learning framework designed for the purpose
of mining gene-disease associations. To tackle the issue of data incompleteness, we ini-
tiate the process by conducting data augmentation on the protein interaction network
through L3-based link prediction algorithms (Fig. 1A). L3-based link prediction algo-
rithms integrate biological motivations into the prediction of protein–protein interac-
tions, surpassing the performance of general-purpose algorithms [15]. Subsequently, the
establishment of a heterogeneous module network (Fig. 1C) unfolds, seamlessly integrat-
ing disease-gene associations, augmented protein interactions, and protein complexes
(Fig. 1B). Within this framework, a sophisticated graph embedding method is devised
to harness the cooperative relationships intrinsic to the heterogeneous module network
(Fig. 1D), subsequently deploying this method to generate candidate genes for each dis-
ease (Fig. 1E). Furthermore, a graph neural network is engineered to glean enhanced
representations by collectively aggregating information from the topological structure
(Fig. 1F). Ultimately, low-dimensional disease and gene embeddings are harnessed for
gene prioritization (Fig. 1G).

Materials and methods

Graph data augmentation based on L3 principle
Even with significant advancements in high-throughput mapping techniques, a consider-
able number of human protein–protein interactions (PPIs) remain unknown compared
to those that have been experimentally documented [16]. Network-based link prediction
algorithms are gaining momentum as valuable computational tools for predicting unde-
tected interactions. Such state-of-the-art algorithms rely on the triadic closure principle,
which assumes that the number of paths of length two between two nodes is correlated
with the likelihood of them also being directly connected. However, the triadic closure
principle inadequately characterize PPIs, thereby failing to guarantee the correctness
and reliability of predictions. Figure 1A illustrates that protein a and protein c share a
path of length 2, indicating a potential interaction based on the triadic closure principle.
PPIs often require complementary interfaces [17, 18]. As a result, protein a and protein
Jia et al. BMC Bioinformatics (2024) 25:214 Page 4 of 14

c exhibit similar interfaces, as illustrated by their identical shapes in Fig. 1A. It is notable
that such an interface does not typically guarantee that protein a and protein c interact
with each other [15].
To address the aforementioned issue, Kovács et al.[15] proposed a novel link predic-
tion predictor based on the L3 principle, positing that proteins linked by multiple paths
of length three are more likely to have a direct link. As shown in Fig. 1A, an additional
interaction partner of protein c (protein d) and protein a have a complementary inter-
face, suggesting a possible direct interaction. Such an interaction can be predicted by
using paths of length three (L3). In this paper, we adopted the L3 principle to perform
data augmentation on the protein interaction network. Three L3 scores are assigned to
each node pair, x and y (Eqs. 1–3)

LCN
3 x, y = axu auv avy (1)
u,v

 
1 1
LRA
  
3 x, y = axu auv avy ku + kv (2)
u,v

 
1 1
LAA
  
3 x, y = axu auv avy logku
+ logkv (3)
u,v

where ku represents the degree of node u while axu is a binary variable. axu = 1 if node
x interacts with node u interacts, otherwise axu = 0. LRA3 And L3 are degree-normal-
AA

ized versions of L3 , derived from the insights obtained from RA (Resource Allocation)
CN

and AA (Adamic-Adar) [19]. When performing data augmentation, taking protein x as

an example, first calculate similarity scores with all remaining nodes (excluding those
already connected to x). Then, select the top l nodes with the highest similarity to x for
3 , L3 , and L3 respectively. The selected node sets are denoted as SCN , SRA, and SAA.
LCN RA AA

Lastly, create edges between x and each node in the set S = S CN ∪ SRA ∪ SAA.

Graph representation for the heterogeneous module network and candidates generation
As illustrated in Fig. 1C, a heterogeneous module network, denoted as G = (V , E), was
constructed by integrating disease-gene associations, augmented protein–protein inter-
actions, and protein complexes (Fig. 1B). In this network, the node set V, consists of
disease and gene nodes, with V = Vd ∪ Vg . And the edge set E, includes disease-gene
associations and protein–protein interactions, E = Edg ∪ Egg . For simplicity, protein
nodes are referred to as gene nodes, and protein interactions are represented as gene
interactions. Certain nodes, such as x and y, exhibit cooperative relationships and belong
to a module, denoted as M1. This can be expressed as x ∈ M1, y ∈ M1, or M1 = {x, y}. M1
is a member of the module set M that comprises of protein complexes.
In this study, Node2vec [20], a prevalent network embedding algorithm, was intro-
duced to extract low-dimensional node representations from the heterogeneous
module network. Firstly, we utilized random walks to generate multiple neighbor
sequences for each node. It should be noted that two types of sequences were gen-
erated for each node: the conventional node sequences Qn and enhanced sequences
Jia et al. BMC Bioinformatics (2024) 25:214 Page 5 of 14

Qm that incorporate both nodes and modules. As depicted in Fig. 1D, the sequence
q1n = g1 → g3 → g5 → d3 . . . is a walk sequence starting from g1 that only contains
node. By replacing gene nodes with their corresponding module numbers (both g1
and g3 belong to M1, so they are both replaced with M1), the sequence q1n can be trans-
formed into q1m = M 1 → M1 → g5 → d3 . . .. Here, q1n ∈ Qn andq1m ∈ Qm . Then, all
the sequences of Qn were treated as texts, where nodes were considered as words,
and the skip-gram model, a typical natural language processing model, was applied
to learn the node embeddings. Similarly, all the sequences of Qm were provided to the
skip-gram model to learn the module embeddings. If a node does not belong to any
module, its node embeddings were used as its module embeddings.
For each disease, we computed cosine similarities between its node embedding and
the embeddings of all gene nodes. Then, we selected the top-k genes with the high-
est similarity as candidates for each disease (Fig. 1E). In the disease gene prediction
stage, we focused only on calculating similarities between each disease and its candi-
date genes, significantly reducing the computational complexity.

Graph neural networks for the heterogeneous module network

A graph neural network was constructed based on the graph representation, aimed
at improving the learning of low-dimensional node representations by aggregating
information from the topological structure. The embedding vectors obtained from
the graph representation served as initial node features for the graph neural net-
work. In the graph neural network architecture (Fig. 1F), a graph attention network
was initially employed to assign different weights to neighbors for updating node
information. Subsequently, two graph convolutional layers were applied to protein
interactions, while two GraphSage layers were used for disease-gene associations.
The heterogeneous module network employed the Graph attention network (GAT)
to compute the hidden states of each node through a self-attention strategy. This can
be defined by Eqs. 4 and 5:

Hi1 = αi,j W 0 Hj0


j∈Ni
(4)

αi,j = softmaxj (eij ) = softmax (LeakyReLU (−

→ T
a [W 0 Hi0 ||W 0 Hj0 ])) (5)
j

where Ni represents the neighborhood set of node i, W 0 is a trainable weight matrix,

Hj0 is the initial features of node j obtained from graph representation, and Hi1 denotes
the embedding vector of node i obtained by GAT. A shared attentional mechanism
a : F × F → (F represents the number of the node features output by the layer) is per-
formed on the nodes to compute attention coefficients eij = a(W 0 Hi0 , W 0 Hj0 ) that
represent the importance of node j ‘s features to node i. αi,j is calculated by normaliz-
ing eij with the softmax function. − →a is a weight vector to parameterize the single-layer
feedforward neural network that forms the attention mechanism a. [W 0 Hi0 ||W 0 Hj0 ]
signifies the concatenation of W 0 Hi0 and W 0 Hj0, and LeakyReLU is the activation
Jia et al. BMC Bioinformatics (2024) 25:214 Page 6 of 14

function. Specifically, Hj0 = [Hjnode ||Hjmodule ], where Hjnode and Hjmodule represent the
node embedding and module embedding of node j obtained from the graph representa-
tion, respectively.
For the subgraph Ggg , the convolution operation was conducted by the graph con-
volutional layer. Graph convolutional layer can be defined as Eq. 6:
 
k+1  1 k k k
Hi =σ cji Hj W + b (6)
j∈Ni

√
where cji = |Nj | × |Ni | , bk is a trainable bias matrix, and W k is a trainable weight


matrix. The activation function σ , set as RELU in this paper, is applied to the layer. Hjk+1
(k ≥ 1) represents the embedding vector of node j in the k + 1th layer, and Hj1 captures
the information of node j obtained by GAT.
GraphSage layer was adopted for the subgraph Gdg . In contrast to the graph convolu-
tional layer that utilizes the full neighborhood set, GraphSage layer samples a specific
proportion of neighbors to aggregate information. The embedding process of Graph-
SAGE is defined by Eqs. 7 and 8:
 
HNk+1
′ = AGG k+1 Hjk , ∀j ∈ Ni′ (7)
i

Hik+1 = σ (W k+1 · [Hik ||HNk+1

i′
]) (8)

where Ni ′ represents a subset from the neighborhood set Ni. The aggregation function,
denoted as AGG k+1, was chosen as the mean aggregator in this study, and hence Graph-
Sage takes the mean over neighbors of node i according to Eq. 7. Different with the graph
convolutional layer, GraphSAGE concatenates the node representation with the mean
aggregation of neighbor nodes as shown in Eq. 8, which avoids node information loss.
The outputs of the various convolutional layers were aggregated to incorporate infor-
mation from all types of edges for each node. In this study, two layers were constructed
for GCN and Graphs sage, which has demonstrated strong performance in prior
research [21, 22]. Our ablation experiments also demonstrated that setting the number
of layers to 2 for both GraphSage and GCN can achieve good results. Please refer to
Sect. "Ablation study" and Supplementary Figs. S1 and S2.

Training and prediction

Denote the embedding of node i obtained from the graph neural network as Hi . To eval-
uate the strength of the association for a disease-gene pair (di, gj), we employed cosine
similarity (Eq. 9) as a measure:

H
 ·H
 i j 

scoreij = (9)
H i H j 
    

where H  i = [Hi ||H node ], H node represents the node embedding obtained from node2vec
i i
 i | is the norm ofH
and |H  i.
During the training phase, negative samples were randomly selected from all uncon-
nected pairs between diseases and genes. Due to the fact that the connected gene-disease
Jia et al. BMC Bioinformatics (2024) 25:214 Page 7 of 14

pairs are significantly less than the unconnected gene-disease pairs, we set the number
of negative samples to be p times the number of positive samples. To learn the param-
eters, the margin loss function was adopted, defined by Eq. 10:
   
Loss yij , 
yij = Max 0,1 −  yij · yij (10)

where yij = scoreij , and yij represents the true relationship between gene node i and dis-
ease node j. Specially, yij = 1 if there exists an association between i and j, otherwise
yij = 0.
During the prediction phase, scores were solely computed for the associations between
each disease and its candidate genes. Afterwards, the candidate genes were ranked for
each disease based on their respective scores.

Results
Datasets
The heterogeneous module network consists of two types of nodes that represent genes
and disease, two types of links corresponding to disease-gene associations and protein–
protein interactions, and one type of modules (protein complexes). The disease-gene
associations and 213,888 protein–protein interactions were downloaded from the litera-
ture [23], which sourced the data from the DisGeNet [24] database. A total of 2822 pro-
tein complexes were collected from Human Protein Reference Database [25].
In accordance with the experimental methodology of the prior research [23], the dis-
ease-gene associations were classified into two distinct groups. The first group, denoted
as the internal dataset, contained 130,820 disease-gene associations involving 13,074 dis-
eases and 8947 genes, which was used for cross validation. The second group comprised
10,066 disease-gene associations involving 1186 diseases and 2552 genes. Termed as the
external dataset, this group was collected from DisGeNet that integrated animal model
data, which was used to assessment the capacity to discover new candidate associations.

Experimental setting
We adopted the experimental settings proposed by Yang et al. [23]. To validate the effec-
tiveness of our method, we conducted a tenfold cross validation on the 130,820 curated
associations. Additionally, we used 10,066 associations from animal model as an external
dataset for each fold. The parameter l in graph data augmentation is set to 10, resulting
in a total of 243,379 newly added interactions. The hyperparameters were tuned with
the help of cross validation. Specially, for the node2vec, we set the window size, the walk
length, the number of walks, the in–out parameter, the embedding size and the itera-
tion number to 5, 64, 10, 0.3, 128 and 10, respectively. For GAT, we set the size of hidden
units for GAT​to (256, 128), and the number of heads in multi-head attention to 2. The
learning rate, epoch number and size of hidden units for GCN and GraphSage were set
to 0.0009, 10 and (128, 64, 8), respectively. Moreover, the number of negative samples
was set to be 50 times ( p = 50) greater than the number of positive samples.
In the experiments, Precision, Recall, F1-score (F1) and Association Precision (AP)
were employed to evaluate the performance of gene prioritization. Denote the true path-
ogenic genes of the disease d in the test set as T(d), and record the top i genes with the
Jia et al. BMC Bioinformatics (2024) 25:214 Page 8 of 14

highest predicted probabilities for the disease d as Pi (d). Precision, Recall, F1-score in
Top@i can be defined as follows:

|T (d) Pi (d)|

1
(11)

Prec = |D| d∈D |Pi (d)|

|T (d) Pi (d)|

1
(12)

Recall = |D| d∈D |T (d)|

2|T (d) Pi (d)|


1
(13)

F1 = |D| d∈D |Pi (d)|+|T (d)|

To assess the overall performance, the association precision (AP) is defined as follows:

 d∈D |
T (d) Pk (d)|
 
AP = min( | P k (d)|,10)
(14)
d∈D

Here, D is the disease set and k is set as the number of true pathogenic genes in the
test for each disease. If the number of pathogenic genes for a certain disease is greater
than 10, then set k as 10. The Eq. 14 imposes restrictions the list length of candidate
genes, focusing solely on the top 10 candidate genes for each disease. This is because the
exploration of gene-disease associations is essentially a ranking problem, and during cell
experiments, animal model studies, and clinical trials, candidates are typically selected
from the top-ranked genes. Additionally, AUC was utilized to evaluate the performance.

Fig. 2 Cross validation performance comparison with state-of-the-art methods on the internal dataset.
A The average F1, Precision and Recall of Top-3 predicted genes. B The average F1, Precision and Recall of
Top-10 predicted genes. C AP performance. D ROC curves for disease gene prediction. Error bars represent
the distribution of tenfold cross validations
Jia et al. BMC Bioinformatics (2024) 25:214 Page 9 of 14

Fig. 3 Performance comparison with state-of-the-art methods on the external dataset. A The average F1,
Precision and Recall of Top-3 predicted genes. B The average F1, Precision and Recall of Top-10 predicted
genes. Error bars represent the distribution of tenfold cross validations

Performance comparisons with state‑of‑the‑art methods

To validate the superiority of our approach, we compared ModulePred with three
state-of-the-art methods including DADA [26], RWR [27], RWRH [28], Dgn2vec [29]
and HerGePred [23]. As depicted in Fig. 2, our approach demonstrated superior per-
formance compared to these competitive methods. In terms of Top@3, ModulePred
exhibited the highest Precision, Recall and F1 (Fig. 2A). Similarly, for Top@10, Mod-
ulePred significantly outperformed the other methods across the three metrics (Fig. 2B).
When evaluating the overall performance using the Association Precision (AP), HerGe-
Pred outperformed the other baseline methods. However, our approach, ModulePred,
showed remarkable improvement over HerGePred, with an increase of approximately 4
percentage points in AP (from 0.259 to 0.306; Fig. 2C) and 7 percentage points in AUC
(from 0.752 to 0.834; Fig. 2D).
To evaluate the capability for discovering new disease genes, we further assess the per-
formance on the external dataset, as depicted in Fig. 3. In the Top@3 scenario, Mod-
ulePred outperformed other methods in terms of F1 and Recall, despite its Precision
being lower than that of RWR and RWRH (Fig. 3A). Moreover, the performance of the
methods in the Top@10 scenario was found to be similar to that in the Top@3 scenario
(Fig. 3B). It is important to note that the performance on the external dataset in Fig. 3
was notably lower than that on the internal dataset in Fig. 2. This discrepancy arises
from the fact that both the external and internal datasets were evaluated using the same
prediction results. For example, assume that disease d is associated with genes g1, g2,
g3, g4 and g5 in the internal dataset, and with genes g6 and g7 in the external dataset. In
a fold of cross-validation, the training set includes two gene-disease associations (d, g 1 )
and (d, g 2 ), while the test set includes (d, g 3 ), (d, g 4 ) and (d, g 5 ). An algorithm predicts
the top 3 candidate genes most likely associated with disease d as g3, g4 and g5. In the
Top@3 scenario, the algorithm achieves 100% precision, recall and F1 score in predicting
disease d . Since the top 3 candidate genes have no intersection with the external dataset,
the algorithm completely fails to discover new genes in the external dataset, leading to a
bias in its performance on the external dataset.
Jia et al. BMC Bioinformatics (2024) 25:214 Page 10 of 14

Ablation study
We compared the proposed ModulePred method with three ablations, namely GNN-M,
­GNN* and GNN. Theses variants were compared as follows:

(1) GNN*-M is the complete ModulePred method which uitlizes the augmented pro-
tein interaction network and applies graph representation with module informa-
tion.
(2) GNN-M is an ablation of ModulePred that applies graph embedding solely on the
original protein interaction network.
(3) GNN* is an ablation of ModulePred that uses the augmented protein interaction
network without modules and performs graph embedding using the traditional
node2vec approach.
(4) GNN is an ablation of ­GNN* that uses the original protein interaction network
without protein complexes.

As depicted in Fig. 4, the incorporation of protein complexes allowed GNN-M to sur-

pass GNN in all the evaluation metrics. Similarly, ­GNN* utilzied the augmented pro-
tein–protein interaction network to investigate the connections between diseases and
genes, resulting in significant notable enhancements across all evaluation metrics com-
pared to GNN. Notably, the impact of data augmentation had a greater impact on the AP
index compared to module information (Fig. 4C). ModulePred, which integrated both

Fig. 4 Cross validation performance comparison with three ablations on the internal dataset. A The average
F1, Precision and Recall of Top-3 predicted genes. B The average F1, Precision and Recall of Top-10 predicted
genes. C AP performance. D ROC curves for disease gene prediction. Error bars represent the distribution of
tenfold cross validations
Jia et al. BMC Bioinformatics (2024) 25:214 Page 11 of 14

Fig. 5 Performance comparison with three ablations on the external dataset. A The average F1, Precision and
Recall of Top-3 predicted genes. B The average F1, Precision and Recall of Top-10 predicted genes. Error bars
represent the distribution of tenfold cross validations

module information and augmented protein interactions, made substantial progress

when compared to ­GNN* and GNN-M (Fig. 4C).
Furthermore, we performed an analysis on the external dataset (Fig. 5), once again
confirming the superiority of ModulePred over three ablations. This reinforced the
potential of our approach in uncovering novel disease-gene associations. Both GNN-M
and ­GNN* consistently exhibited better performance than GNN. However, GNN-M out-
performed better than ­GNN* on the external dataset, showcasing a deviation from their
performance on the internal dataset.
We further conducted three additional ablation experiments. Figure S1 indicates that
the network structure of ModulePred (utilizing GAT for processing heterogeneous net-
works, GraphSage for processing gene-disease associations, and GCN for processing
protein–protein interactions) can achieve good performance. Figure S2, suggests that
setting the number of GAT layers to 1, GraphSage to 2 and GCN to 2 in ModulePred is
an optimal parameter configuration. Moreover, Figure S3 demonstrates that setting the
number of l in graph data augmentation to 10 can achieve optimal performance.

Table 1 Top 10 predicted genes for IPAH

Rank Gene Reference

1 MIR204 PMID: 30,854,934

2 CBLN2 PMID: 27,770,446
3 OTSC1 NA
4 EIF2AK4 PMID: 31,711,431
5 ENG PMID: 30,312,106
6 PYCR1 NA
7 RTEL1 PMID: 30,523,160
8 LBR NA
9 B3GAT3 NA
10 TGFBR3 PMID: 11,282,888
Jia et al. BMC Bioinformatics (2024) 25:214 Page 12 of 14

Table 2 Top 10 predicted genes for Hypothyroidism

Rank Gene Reference

1 LHX3 PMID: 12,244,277

2 OTX2 PMID: 26,416,826
3 GALE NA
4 MAGEL2 PMID: 33,570,896
5 BRAF PMID: 21,512,141
6 GLI2 PMID: 25,484,916
7 FANCB PMID: 28,588,452
8 CDKN1C NA
9 NDST1 NA
10 PAH NA

Case study
To further elucidate the biological insights of our approach, we conducted two case
studies in order to identify disease genes related to hypothyroidism and Idiopathic Pul-
monary Arterial Hypertension (IPAH). The predicted genes were ranked based on their
scores (refer to Eq. 9 for details). Furthermore, we manually searched published biomed-
ical literature to obtain final confirmations.
IPAH is a progressive and potentially life-threatening condition characterized by
elevated blood pressure in the pulmonary arteries without any discernible underlying
cause, requiring thorough investigation and management from a medical perspective
[17]. Among the top 10 genes predicted by ModulePred (Table 1), an impressive 6 asso-
ciations were substantiated by previous publications, supported by their correspond-
ing PubMed Unique Identifier (PMID). For instance, the top-ranked gene MIR204 has
been reported to exhibit abnormal expression in relation to the onset and progression of
IPAH [30].
Hypothyroidism is a multifaceted endocrine disorder characterized by diminished
production or action of thyroid hormones, resulting in a variety of physiological dis-
ruptions that necessitate investigation and management from an endocrinological per-
spective. Recent studies have identified several genes associated with hypothyroidism
[31–33]. As presented in Table 2, our ModulePred achieved high prediction accuracy
rates of 100%, 80%, 86% for the top 2, top 5 and top 7 genes, respectively. For instance,
OTX2 Mutations have been linked to developmental abnormalities in both the central
nervous system and the thyroid, resulting in hypothyroidism [34]. Similarly, defects
in GLI2 can disrupt normal thyroid development and function, potentially leading to
reduce thyroid hormone levels [35].

Conclusion
In this article, a deep learning framework called ModulePred is presented for predicting
disease-gene associations. ModulePred achieves competitive predictive performance by
employing graph augmentation on the protein interaction network and graph embed-
ding for the heterogeneous module network. Experimental results on the DisGeNet
dataset substantiate the efficacy of ModulePred in discovering disease-gene associations.
Furthermore, the ablation study highlights the greater impact of graph augmentation
Jia et al. BMC Bioinformatics (2024) 25:214 Page 13 of 14

on the performance of ModulePred compared to the graph embedding for the module
network.

Abbreviations
SNPs Single nucleotide polymorphisms
GCN Graph convolutional network
PPI Protein–Protein interaction
GAT​ Graph attention network
F1 F1-score
AP Association precision
IPAH Idiopathic pulmonary arterial hypertension
PMID PubMed unique identifier

Supplementary Information
The online version contains supplementary material available at [Link]

Supplementary Material 1.

Acknowledgements
We thank Mr. Yi Zhao from Qingdao University for the support of computing resources.

Author contributions
S.W. and X.J. conceived the idea. X.J., W.L. and J.L. implemented the algorithm and codes. X.J., W.L., H.S. and J.X. performed
the analysis. W.L. and J.X. prepared figures. S.W. and X.S. wrote the manuscript. All contributed the proofread.

Funding
XS acknowledges support of Grant No. 2021YFF0704500 from National Key Research and Development Program of
China, Grant No. 32070086 from National Natural Science Foundation of China, Shandong Province Youth Entrepre-
neurial Talent Introduction and Training Program, and Shandong Province Taishan Scholars Youth Experts Program. SW
acknowledges support of Grant No. ZR2019PF012 from Shandong Provincial Natural Science Foundation of China.

Availability of data and materials

All datasets and code in this work are available at [Link] All other relevant data is
available upon request.

Declarations
Ethics approval and consent to participate
Not applicable.

Consent for publication

Not applicable.

Competing interests
The authors declare that they have no competing interests.

Received: 14 March 2024 Accepted: 12 June 2024

References
1. Yoon S, Nguyen HCT, Yoo YJ, et al. Efficient pathway enrichment and network analysis of GWAS summary data using
GSA-SNP2. Nucleic Acids Res. 2018;46(10):160.
2. Ata SK, Wu M, Fang Y, et al. Recent advances in network-based methods for disease gene prediction. Brief Bioinform.
2020. [Link]
3. Ghiassian SD, Menche J, Barabasi AL. A DIseAse MOdule Detection (DIAMOnD) algorithm derived from a sys-
tematic analysis of connectivity patterns of disease proteins in the human interactome. PLoS Comput Biol.
2015;11(4):e1004120.
4. Himmelstein DS, Lizee A, Hessler C, et al. Systematic integration of biomedical knowledge prioritizes drugs for
repurposing. Elife. 2017. [Link]
5. Valdeolivas A, Tichit L, Navarro C, et al. Random walk with restart on multiplex and heterogeneous biological net-
works. Bioinformatics. 2019;35(3):497–505.
6. Lin CH, Konecki DM, Liu M, et al. Multimodal network diffusion predicts future disease-gene-chemical associations.
Bioinformatics. 2019;35(9):1536–43.
7. Jiang B, Kloster K, Gleich DF, et al. AptRank: an adaptive PageRank model for protein function prediction on bi-
relational graph. Bioinformatics. 2017;33(12):1829–36.
Jia et al. BMC Bioinformatics (2024) 25:214 Page 14 of 14

8. Lotfi Shahreza M, Ghadiri N, Mousavi SR, et al. A review of network-based approaches to drug repositioning. Brief
Bioinform. 2018;19(5):878–92.
9. Jowkar GH, Mansoori EG. Perceptron ensemble of graph-based positive-unlabeled learning for disease gene identi-
fication. Comput Biol Chem. 2016;64:263–70.
10. Chen X, Yan CC, Zhang X, et al. Long non-coding RNAs and complex diseases: from experimental results to compu-
tational models. Brief Bioinform. 2017;18(4):558–76.
11. Li Y, Patra JC. Integration of multiple data sources to prioritize candidate genes using discounted rating system. BMC
Bioinform. 2010;11(Suppl 1):S20.
12. Yang K, Wang R, Liu G, et al. HerGePred: heterogeneous network embedding representation for disease gene pre-
diction. IEEE J Biomed Health Inform. 2019;23(4):1805–15.
13. Cinaglia P, Cannataro M. Identifying candidate gene-disease associations via graph neural networks. Entropy (Basel).
2023;25(6):909.
14. Zhang J, Zhong C, Huang Y, et al. A method for identifying protein complexes with the features of joint co-localiza-
tion and joint co-expression in static PPI networks. Comput Biol Med. 2019;111:103333.
15. Kovacs IA, Luck K, Spirohn K, et al. Network-based prediction of protein interactions. Nat Commun. 2019;10(1):1240.
16. Luck K, Sheynkman GM, Zhang I, et al. Proteome-scale human interactomics. Trends Biochem Sci. 2017;42(5):342–54.
17. Keskin O, Tuncbag N, Gursoy A. Predicting protein-protein interactions from the molecular to the proteome level.
Chem Rev. 2016;116(8):4884–909.
18. Szilagyi A, Grimm V, Arakaki AK, et al. Prediction of physical protein-protein interactions. Phys Biol. 2005;2(2):S1-16.
19. Lu L, Zhou T. Link prediction in complex networks: a survey. Physica A: Stat Mech Appl. 2010;390(6):1150–70.
20. Grover A, Leskovec J. node2vec: scalable feature learning for networks. KDD. 2016;2016:855–64.
21. Kipf TN, Welling M. Semi-supervised classification with graph convolutional networks. arXiv:1609.​02907. 2016.
22. Hamilton W, Ying Z, Leskovec J. Inductive representation learning on large graphs. Adv Neural Info Proc Syst
2017;30.
23. Yang K, Wang R, Liu G, et al. HerGePred: heterogeneous network embedding representation for disease gene pre-
diction. IEEE J Biomed Health Inform. 2019;23(4):1805–15.
24. Pinero J, Bravo A, Queralt-Rosinach N, et al. DisGeNET: a comprehensive platform integrating information on human
disease-associated genes and variants. Nucleic Acids Res. 2017;45(11):D833–9.
25. Chan W. The UniProt Knowledgebase (UniProtKB): a freely accessible, comprehensive and expertly curated protein
sequence database. Genet Res. 2010;92(1):78–79.
26. Erten S, Bebek G, Ewing RM, et al. DA DA: degree-aware algorithms for network-based disease gene prioritization.
BioData Min. 2011;4(1):1–20.
27. Fernández P. Google’s pagerank and beyond: the science of search engine rankings. Math Intell. 2008;30(1):68–9.
28. Cao Z, Wei F, Dong L, et al. Ranking with recursive neural networks and its application to multi-document summari-
zation. Proceed AAAI Conf Artif Intell. 2015. [Link]
29. Liu Y, Guo Y, Liu X, et al. Pathogenic gene prediction based on network embedding. Brief Bioinform.
2021;22(4):bbaa353.
30. Estephan LE, Genuardi MV, Kosanovich CM, et al. Distinct plasma gradients of microRNA-204 in the pul-
monary circulation of patients suffering from WHO Groups I and II pulmonary hypertension. Pulm Circ.
2019;9(2):2045894019840646.
31. Hwangbo Y, Park YJ. Genome-wide association studies of autoimmune thyroid diseases, thyroid function, and
thyroid cancer. Endocrinol Metab (Seoul). 2018;33(2):175–84.
32. Teumer A, Chaker L, Groeneweg S, et al. Genome-wide analyses identify a role for SLC17A4 and AADAT in thyroid
hormone regulation. Nat Commun. 2018;9(1):4455.
33. Stoupa A, Adam F, Kariyawasam D, et al. TUBB1 mutations cause thyroid dysgenesis associated with abnormal
platelet physiology. EMBO Mol Med. 2018. [Link]
34. Schoenmakers N, Alatzoglou KS, Chatterjee VK, et al. Recent advances in central congenital hypothyroidism. J Endo-
crinol. 2015;227(3):R51–71.
35. Ma D, Marion R, Punjabi NP, et al. A de novo 10.79 Mb interstitial deletion at 2q13q14.2 involving PAX8 causing
hypothyroidism and mullerian agenesis: a novel case report and literature review. Mol Cytogenet. 2014;7(1):85.

Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.