Table 2 Average network distance of drug targets without and with known side effects used in the treatment of colorectal cancer and type 2 diabetes from the disease-associated proteins.

From: Targets of drugs are generally and targets of drugs having side effects are specifically good spreaders of human interactome perturbations

Protein group

Average network distance from disease-related proteins(edges)

24 drug targets without known side effects used in the treatment of colorectal cancer

2.528

3 drug targets with known side effects used in the treatment of colorectal cancer

2.389

14 drug targets without known side effects used in the treatment of type 2 diabetes

3.250*

25 drug targets with known side effects used in the treatment of type 2 diabetes

3.234**

  1. *This value is significantly greater than the average network distance of drug targets without known side effects in colorectal cancer (p = 1.062e-05). Statistical analysis was performed using the Welch (Student’s) two sample t-test function of the R package56.
  2. **This value is significantly greater than the average network distance of drug targets with known side effects in colorectal cancer (p = 0.005441). Statistical analysis was performed using the Welch (Student’s) two sample t-test function of the R package56.
  3. The table shows the arithmetic mean of the average network distance between drug targets (with and without known side effects used in the treatment of colorectal cancer and type 2 diabetes) and the proteins related to the respective disease (results were very similar, if instead of arithmetic means we used the medians; data not shown). The total number of colorectal cancer- and diabetes-related proteins in the human interactome were 18 and 14, respectively. Average network distances were calculated as shortest paths using the Pajek programme58. Proteins were labelled by their UniProt ID54. Human interactome containing 12,439 proteins and 174,666 edges was built from the STRING database46, 1,726 human drug targets were obtained from the DrugBank database47 and 99,423 drug-side effect pairs were taken from the SIDER database2. Colorectal cancer- and type 2 diabetes-related proteins were obtained from the Cancer Gene Census database48 and from the article of Parchwani et al.49, respectively. We used the mean values and the t-test because of the near-normal distribution of the average network distances.
Back to article page