4.6 Aging Co-Expression Network: AgingNet
Last updated
Last updated
AgingNet consists of a co-expression network created through ageCollapsed, where 3569 transcripts were selected to constitute its nodes and the DTW-MIC similarity its interrelationships. These transcripts were selected by correlating their expression profile with age, in which only those with correlations greater than 0.35 were selected (linear correlation: |Rho| > 0.35 and non-linear correlation: MIC > 0.35). Next, groups of very similar transcripts were identified through clustering by the hdbscan algorithm, finding 3 large co-expression modules, named M0, M1 and M2. M0 is a “residual” module, in which it represents transcripts that are not members of any cluster, being disregarded in further analysis.
The remaining modules, M1 and M2, are made up of 1154 and 231 transcripts, respectively, and the most important are illustrated in Figures 4.9 A and B. if only the 100 most forcefully transcribed. It is important to note that 5 AgingGenes are present among the 100 most important transcripts of the two modules: the GLS, TERF2 and MYC genes present in the M1 module and the GYPC and NGLY1 genes present in the M2 module. In addition, 9 transcripts among them are also present in the GenAge database, being the RAE1, ARNTL, MYC and TERF2 genes present in the M1 module, and HDAC3, PRKDC, VCP, SP1 and FOXO1 in the M2 module.
Module M1 (Figure 4.16 A) is mostly permeated with transcripts linearly correlated with age, with 924 transcripts with negative correlations and 247 positive ones. The remaining 42 transcripts have nonlinear correlations. The 10 most important transcripts in this network are RABEPK, CHD1L, POGK, RAE1, LMO4, CNOT6, TRIAP1, MAPKAPK5-AS1, ARNTL and RNF115, all with negative correlations. In addition, this module also has two non-protein-coding transcripts: MAPKAPK5-AS1 and GC01M033423.
Module M2 also has most transcripts with negative correlations (162 transcripts) and 55 transcripts with positive correlations and 23 nonlinear. The 10 most important transcripts of this network are WDR48, XKR8, MRPL24, TM7SF3, BTBD1, HDAC3, PRKDC, RAP1GDS1, COX16 and API5, all negatively correlated with age.
When analyzing the general expression profile of the transcripts that constitute each module, 4 transcripts subgroups were identified in each module with distinct expression profiles, resulting in 8 sub-modules. The sub-modules of M1 and M2 were named M1S1, …, M1S4 and M2S1,..., M2S4, respectively. Next, each sub-module was summarized through the median expression of its transcripts for each age, and its general expression profile was constructed, illustrated in Figures 4.10 A and B. the ages where changes occurred, in the same way as for the AgingGenes.
In the pathway enrichment analysis, two large intra-connected structures were found for module M1 (Figure 4.11 A and B), while for module M2 we found 6 smaller and more disconnected structures (Figure 4.11 C).