• 2019-07
  • 2019-08
  • 2019-09
  • 2019-10
  • 2019-11
  • 2020-03
  • 2020-07
  • 2020-08
  • 2021-03
  • br Association of di erentially


    2.2. Association of differentially expressed miRNAs and patient prognosis
    Profiles of miRNAs with differential Tunicamycin were log2 trans-formed. The prognostic values of all differentially expressed miRNAs were estimated using the log-rank analysis and Kaplan-Meier curve. MiRNAs that presented significant association with patient survival  Gene 699 (2019) 125–134
    performance were referred to as prognostic miRNAs.
    2.3. Target gene prediction of prognostic miRNA
    Online analysis instruments, including miWalk (http://mirwalk., PicTar (, miRDB ( and TargetScan (http://, were used to predict the prognostic target genes of miRNAs. To improve the reliability of the bioinformatic ana-lysis, the overlapping target genes were determined using Venn dia-gram. Next, the KOBAS online analysis database (http://kobas.cbi.pku. was used to perform the KGEE disease and KEGG pathway analyses on the overlapping genes.
    2.4. Enriched ontology analysis
    The Metascape database ( main/step1) was used to perform GO annotations of these target genes. A representative term subset was selected from the cluster, and converted in a network layout. In particular, circle nodes were used to represent each term, the sizes of which proportionally grew with the number of input genes coinciding with the corresponding terms, and the color of which represented the identity of the cluster it was selected from. In other words, same-color circle nodes represented terms from the same cluster. An edge was used to link the terms that presented a similarity score higher than 0.3 (the score of similarity was represented by the thickness of this edge). Cytoscape was used to visualize this network, with bundled edges and force-directed layout to ensure its clarity. A specific description label was attached to every term selected from each cluster.
    2.5. Module screening from the PPI network
    For each target gene, PPI enrichment analysis was performed using BioGrid (Stark et al., 2006), InWeb IM (Li et al., 2017), OmniPath (Li et al., 2017). The resultant network contained the subset of proteins that formed physical interactions with at least another list member. The sizes of nodes proportionally grew with the number of interactions. The Molecular Complex Detection (MCODE) algorithm was then applied to this network to identify neighborhoods of densely connected proteins. A unique color was assigned to every MCODE network. GO enrichment analysis was performed on every MCODE network to provide meaning to all the components of the network.
    2.6. Statistical analysis
    Mean ± standard deviation (SD) was used to express analytical data. Unpaired t-test was conducted for analyzing the miRNA expres-sion levels in GC and normal tissues. t-Tests were performed for the assessing of the relationship of clinical characteristics with miRNAs expression. Univariate/multivariate Cox proportional hazard regression analysis and Kaplan-Meier survival analysis were conducted to compare the high-level and low-level miRNAs. The IBM SPSS software version 20.0 (IBM Corp., NY, USA) was used to analyze on statistical data. A nomogram was built on the basis of multivariate analysis results and selected miRNAs use of rms package in R version 3.3.3 ( The calculated C-index was used to measure nomogram performance, which was assessed by comparing between the survival probability predicted by the nomogram and the actual survival prob-ability obtained using Kaplan-Meier estimates. The predicting perfor-mance of the nomogram was evaluated by comparing the C-index of the established nomogram and other staging systems (Huitzil-Melendez et al., 2010). For validating this established nomogram, the total score of all patients in the validation cohort was calculated in accordance with this nomogram, which was then used as a factor for Cox regression analysis on the cohort. Finally, the calibration curve and C-index were