Essential Gene Knowledge Graph

    Based on specificity of gene essentiality in various cancer types, we developed a new strategy to screen targets and identify candidate drugs. Specifically, we defined GeneSpecificityScore (e score) for reflecting the specific essentiality of genes in cancer types. We then ranked the genes by GeneSpecificityScore and selected the top ones as drug targets. Secondly, combining target-drug interaction databases TTD (https://db.idrblab.net/ttd/) and DGIdb (https://www.dgidb.org/) with the research/marketing status of existing drugs/compounds, potential drugs against cancer-type-specific targets were picked out.

    We verified the feasibility of our strategy through experiment. All of the seven identified drugs with no prior anti-cancer evidences were validated through inhibiting experiment in 11 cell lines of lung adenocarcinoma. Furthermore, lower inhibiting rates in 10 normal cell lines were observed for all the drugs. Colchicine was found to have the highest inhibiting effect on lung adenocarcinoma cells. We also proposed the concept of orthogonal drug targets to summate the effect of multiple drugs and verified the anti-lung cancer effects of progesterone and rosiglitazone in vitro. The results of cell level showed that using both drugs simultaneously could achieve improved chemotherapy effects while greatly reducing the damage to normal cells. Our proposed systematic strategy could be used to screen type-specific targets and find more therapies for other cancers.

    The online service was established to graphicly display the relationship between cancer types, essential genes, and drugs. In addition, the results of our experiments were graphically displayed. And for external validation, the results of cellular experiments from DepMap were linked to.

Cancer Type Number of Cell Lines
Acute_Myeloid_Leukemia 2
B_Cell_Non_Hodgkin`s_Lymphoma 5
Biliary_Tract_Carcinoma 1
Blood_Non_Cancerous 2
Breast_Carcinoma 25
Colorectal_Carcinoma 30
Endometrial_Carcinoma 7
Esophageal_Adenocarcinoma 7
Esophageal_Squamous_Cell_Carcinoma 19
Ewing`s_Sarcoma 10
Eye_Non_Cancerous 1
Gastric_Carcinoma 14
Glioblastoma 24
Head_and_Neck_Carcinoma 16
Kidney_Carcinoma 3
Low_Grade_Glioma 9
Lung_Adenocarcinoma 20
Melanoma 4
Mesothelioma 1
Neuroblastoma 17
Non-Small_Cell_Lung_Carcinoma 9
Oral_Cavity_Carcinoma 18
Osteosarcoma 2
Other_Central_Nervous_System_Carcinomas 1
Other_Endometrial_Cancers 1
Other_Lung_Carcinomas 1
Ovarian_Carcinoma 31
Pancreatic_Carcinoma 23
Plasma_Cell_Myeloma 5
Prostate_Carcinoma 3
Soft_Tissue_Carcinomas 1
Squamous_Cell_Lung_Carcinoma 11

Tips:

    The table on the left lists the number of cell lines for each cancer type involved in this study. We assess the specific necessity of a gene in a cancer type based on its necessity across the cell lines of that type. When the number of cell lines is too small, it may lead to inaccuracies in our assessment of the gene's necessity. Therefore, our main focus was on cancer types with 5 or more cell lines, and data for cancer types with fewer than 5 cell lines are included in the "Supplement" interface and should be referred to with higher caution. In the future, we aim to enhance our work using richer and more comprehensive cancer cell line data.