Systematic identification of cancer-type specific drugs based on essential genes and validations in lung adenocarcinoma

doi:10.21203/rs.3.rs-5021281/v1

Download PDF

Research Article

Systematic identification of cancer-type specific drugs based on essential genes and validations in lung adenocarcinoma

https://doi.org/10.21203/rs.3.rs-5021281/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background Depicting a global landscape of essential gene-targeting drugs would provide more opportunities for cancer therapy. Essential genes have been determined in many cancer cell lines. However, systematic investigation on drugs targeting essential genes still has not been reported.

Methods We suppose that drugs targeting cancer-type specific essential genes would generally have less toxicity than those targeting pan-caner essential genes. A scoring function-based strategy was developed to identify cancer-type specific targets and drugs. The EssentialitySpecificityScore ranked the essential genes in 19 cancer types, and 1151 top genes were identified as cancer-type specific targets. Combining target-drug interaction database records with drug research/marketing status, 370 cancer-type specific drugs were identified, bound to 100 out of all identified targets. Cell-inhibiting experiments evaluated the seven chosen drugs’ toxicity on normal and cancer cells. Student’s t-test was used to compare the difference in inhibiting rates. Pearson’s correlation analysis measures the association of targets’ pan-cancer essentialities and drugs’ inhibiting rates on normal cells.

Results Profiles of applied cancer types of identified targets and drugs illustrate the effectiveness of the scoring strategy and most of them apply to cancer types less than 10. Seven drugs with no previous anti-cancer evidence were validated in 11 lung adenocarcinoma cell lines, and lower inhibition rates (from 9.4% to 44.0%) were observed in 10 normal cell lines. This difference is statistically significant (Student’s t-test, p = 0.0001). We also found that the aggregative essentiality values of potentially bound targets are correlated (Pearson's correlation analysis, p = 0.062) with the inhibiting rate of seven tested drugs on normal cells, further confirming the rationality of our supposition.

Conclusions Our work, for the first time, systematically identified drugs targeting cancer-type-specific essential genes and validated the safety of such drugs to normal cells compared to lung adenocarcinoma cells. The EGKG forms a computational basis to uncover essential gene targets and drugs for specific cancer types. Also, our cell experimental result suggests that combining drugs with orthogonal essentiality may be an alternative way to improve anti-cancer effects while maintaining biocompatibility.

cancer-type specific drugs

essential genes

EssentialitySpecificityScore

lung adenocarcinoma

cell inhibition rate.

The 90% failure rate in the development of cancer drugs is due to a lack of efficacy, and effective anti-cancer drug entities need to be developed [1,2]. One of the most effective means of cancer treatment is targeted therapy [3,4], which uses a gene with a definite effect as a drug target [5]. Functional genomics approaches can guide the mining of novel targets [6] from tumor genomic data to develop personalized anti-cancer drugs [7,8]. However, the abundance of genomic information [7], limited reliable target identification [9], lack of clinical efficacy [1], high toxicity, and side effects of anticancer drugs [10] are still obstacles to developing personalized cancer drugs [11]. Identifying essential genes provides the basis for targeted cancer therapy [12]. Essential genes support vital cell functions [13] and can serve as excellent target genes [14,15]. Prioritizing essential genes as targets can provide more reliable choices, increase the feasibility of personalized cancer therapy [16], and facilitate the development of novel cancer drugs. Due to the high mortality rate and the lack of effective drugs [17], the reuse of existing drugs has become a hot topic of research in recent years [18], which is expected to overcome the above limitations that hinder the development of cancer therapy.

At present, many studies have focused on drug target screening [19] and vaccine design [20] around essential genes. Bacterial essential genes have been well studied and widely used in the identification of antimicrobial target genes [21]. For human cell lines, a variety of essential gene recognition models have also emerged. In 2019, the Sanger Center identified essential genes in 325 cell lines using CRISPR high-throughput knockout technology, enabling exploiting these genes for drug screenings [22]. In 2021, Liu’s group from Tongji University analyzed the synthetic lethality of cancer cell lines and established a map of cancer-type therapeutic drugs based on the synthetic lethal gene pairs [23]. However, until now, there is still a lack of systematic studies for cancer therapeutics targeting essential genes [24]. The drug targets need to be highly specific to kill tumor cells without affecting normal cells [25]. Cancer-type specific targets usually are usually identified by their abnormal expression level caused by coding gene mutation or epigenetic modification of promoter region [26–30]. Behan and Colleagues connected cancer-type specific targets with essential genes [22]. Cancer-type specific essential genes, functions of which are only necessary for a certain cancer type or most cell lines of that cancer type and are non-essential in most cell lines of the other cancer types, constitute ideal drug targets [31,32]. Now it is urgent to depict a global landscape of drugs targeting cancer-type specific essential genes. Good antineoplastic drugs kill tumors with little or no toxicity to normal cells [24]. It is reasonable that drugs targeting essential genes specific to a certain cancer type would meet the above requirements in principle.

Accordingly, we proposed a prioritization strategy of antitumor drugs based on essential genes. This systematic strategy contains three steps and involves three scores, expressed in two scoring functions. After identifying cancer-type specific drugs, an online service (http://guolab.whu.edu.cn/egkg/index.php) was established to display the drug prioritization map graphically. Then, we indeed observed the lower toxicity of the top prioritizing drugs to normal cells than cancer cells for lung cancer in vitro. Therefore, our work obtained the complete list of drugs targeting cancer-type specific essential genes and validated the relatively higher safety of such types of drugs at the cell level. Also, we experimentally observed combining drugs with orthogonal essentiality has a stronger anti-cancer effect than the use alone, whereas the biocompatibility is superior. All these computational and experimental results would have insights and implications for cancer targeted therapy.

Building of EGKG

The overall architecture of EGKG and the following analysis are shown in Fig. 1. EGKG consists of three entities: genes, drugs, and cancer types. They are associated with three relationships: (1) relationships between different cancer types and essential genes, (2) relationships between drugs and their target genes, and (3) relationships between drugs and predicted cancer type, which is assigned by the affiliated cancer type of target gene.

Entity and relationship establishment of EGKG

The data sources and processing of EGKG are described in detail below. The information on whether a gene is essential in a cell line was obtained from Sanger's lab [22] (https://score.depmap.sanger.ac.uk/), which helped us to construct the relationship between genes and cancer types through the relationship between genes and cancer cell lines. Gene-drug records, that is the information on whether a gene is a potential target of a drug provided by TTD (https://db.idrblab.net/ttd/) [31] and DGIdb (https://dgidb.org/) [33]. The specific source and number of gene-drug records are shown in Table S1. Then, we identified drug candidates for each cancer type based on the relationship between genes and cancer types, gene-drug records, as well as the research and marketing (R&M) status of drugs. As for the gene-drug interaction, we do not use any quantitative score and only mark it as yes or no based on whether there is a supporting database source.

From Sanger's lab, we obtained 7470 genes that were essential or not in 325 cell lines. EssentialitySpecificityScore (e score) was calculated according to the formula given below. Genes with no gene-drug records were removed, and then the e score was sorted in descending order. Genes with top rankings were selected to be matched with drugs based on gene-drug records. Meanwhile, in order to integrate the drug R&M status and target gene type, DrugScore was calculated as described below and drugs with DrugScore less than 0.8 were filtered out.

Due to drug name inconsistency in TTD [31] and DGIdb [33], when integrating the data, we standardized the drug names with the information in Drugbank (https://go.drugbank.com/) [34]. At the same time, we also supplemented the R&M status information of some drugs using Drugbank [34]. In addition, Drugbank [34] also provided us with some clinical indications of the drugs, which helped us a lot in our research.

Drug reorientation scoring model

In this study, we developed an effective drug repurposing scoring scheme by integrating two core scoring functions, EssentialitySpecificityScore (Fig. 1A) and Drug status score (DrugScore) (Fig. 1C).

EssentialitySpecificityScore is calculated as follows:

Ncl: The number of cell lines for which a gene is essential in a cancer type;

Ntcl: Total number of cell lines of this cancer type;

Nocl: The number of cell lines for which a gene is essential in other cancer types (in the case of lung adenocarcinoma, this number denotes the cell lines for which a gene is essential in cancer types other than lung adenocarcinoma);

Ntocl: Total number of cell lines for other cancer types.

Ideally, the higher the e score, the greater the difference between the essentiality of this gene in a given cancer type and that in normal cells. To guarantee the statistical significance of the n value, we only retain these cancer types with five or more cell lines. That is to say, we identified cancer-type specific targets for 19 cancer types.

Secondly, referring to the strategy applied in SLKG, for the purpose of integrating drug R&M status and target gene type, drug scoring is defined as [23]:

Table 1

Scoring criteria for *DrugScore.*
Development status of the drug/compound	Type of target gene [23]	DrugScore value
Approved	Direct target	1
Approved	Others	0.8
Clinical trials	Direct target	0.6
Clinical trials	Others	0.4
Experimental	Direct target	0.2
Experimental	Others	0.1
Others	Direct target	0.05
Others	Others	0.01

The detailed scores corresponding to specific status and drug target type are listed in Table 1. DrugScore is defined as the maximum of the set of values of Pre-computed DScore values to avoid multiple sources leading to multiple drug R&M statuses. A drug may act on multiple target genes and this may cause one drug to have multiple Target-Gene-Type [18]. In addition, we also unified the drug names in the TTD [31] database and DGI [33] database according to Drugbank’s records [34]. Matching the index ID of all three databases (pubchem_cid/pubchem_sid/chembl_id) at the same time makes it easy to query for drug-specific information and delete drugs that do not have these three IDs (more specific information cannot be obtained).

In order to determine the threshold scores for the best repurposing drug candidates, we further examined the probability and cumulative distribution curves of the EssentialitySpecificityScore and DrugScore. In the probability distribution curves (data not shown), the EssentialitySpecificityScore is enriched in the regions with low scores. Therefore, the thresholds for the two scoring functions are as follows:

In total, we calculated the three scores for 10,376 drugs derived from public databases and identified 370 drugs that are potentially effective in at least one cancer type. We have used three scores affiliated with two functions to identify potential cancer-type specific drugs. After finishing the identification process, we will rank the identified drugs in ascending order of the e score.

EGKG web server development

The EGKG web server can be found at http://guolab.whu.edu.cn/egkg/index.php. It was developed by PHP + JavaScript, and the visualization part is based on the D3.js implementation. D3.js provides a rich set of charts and visualization components, allowing us to create a very user-friendly and interactive data visualization interface. In the “Target” module of EGKG, the chosen target genes for each cancer type involved in this article can be browsed (fig. S7A). In the “Drug” module, users can browse the corresponding knowledge graph based on drug, cancer type, and gene, respectively (fig. S7B, S7C, S7D). If one target has many bound drugs, we will show those drugs with direct (experimental) proof of gene-drug interaction before those with computational proofs. In addition, for drugs that have been validated by in vitro cell experiments, the experimental data are also presented at EGKG (fig. S7E, S7F). The effect of the drug on cancer cell lines and normal cell lines is distinguished by red and blue (fig. S7E). At the same time, the intensity of the effect is also expressed by the shade of color: the darker the red, the better the inhibitory effect of the drug on cancer cell lines, and the darker the blue, the less harmful the drug has on normal cells. The line chart (fig. S7F) shows the trend of drug action on cell lines at different concentrations. To better validate the feasibility of our method, if the identified drugs for each cancer type have cellular experiments recorded in the DepMap database, we provide the corresponding links for them in the “External validation” interface. The data provided by EGKG can be downloaded for further study. The functions of each module of EGKG are described in its “Help” interface.

In vitro experimental section

Materials

All chemicals purchased from commercial suppliers were used without further purification. Apremilast (Purity ≥ 99%), hydrochlorothiazide (≥ 99%), and progesterone (≥ 99%) were purchased from Aladdin. Dexamethasone (98%), ibuprofen (≥ 98%), and colchicine (98%) were purchased from Meryer. Rosiglitazone (99%) was purchased from Macklin. RPMI-modified medium and high-glucose DMEM medium were obtained from Cytiva. 1× phosphate-buffered saline (PBS), 1% antibiotic solution (penicillin-streptomycin, 10,000 U·mL^− 1), and trypsin solution were provided by Biosharp. RPMI 1640 (1X), the cell counting kit-8 (CCK-8), and DMSO were supplied by Meilunbio. Fetal bovine serum (FBS) was obtained from Gibco Life Technologies (USA).

Cell culture

A549, 293T, HUVEC, and Beas-2b were provided by the Chinese Type Culture Collection (Wuhan University). GES-1, NCL-H1993, NCL-H2087, NCL-H3122, NCL-H1650, NCL-H1755, NCL-H1944, NCL-H1975, NCL-H23, NCL-H358, LX-2, hTERT-HPNE, AC16, ARPE-19, SW-1573, MCF 10A, and WPMY-1 were purchased from the Global Bioresource Center (ATCC). GES-1, NCL-H1993, NCL-H2087, NCL-H3122, NCL-H1650, NCL-H1755, NCL-H1944, NCL-H1975, NCL-H23, and NCL-H358 were cultured with RPMI 1640 medium. Beas-2b, HUVEC, LX-2, hTERT-HPNE, AC16, ARPE-19, 293T, SW-1573, and WPMY-1 were cultured with high-glucose DMEM medium. MCF 10A was cultured with DMEM medium supplemented with 5% Horse serum, 0.01 mg·mL^–1 Insulin, 20 ng·mL^–1 EGF, 0.5 µg·mL^–1 Hydrocortisone, 100 ng·mL^–1 Cholera toxin and 1% antibiotic solution (penicillin-streptomycin, 10,000 U·mL^− 1). A549 were cultured with RPMI medium. RPMI medium, high-glucose DMEM medium, and RPMI 1640 medium contained 1% antibiotic solution (penicillin-streptomycin, 10,000 U·mL^− 1) and 10% fetal bovine serum (FBS). These cells were incubated in a humidified incubator of 5% CO₂ and 95% air at 37 ℃.

Cytotoxicity assessment in vitro

The cell cytotoxicity in vitro was monitored with the CCK-8 array. Respectively, 293T, GES-1, HUVEC, Beas-2b, LX-2, hTERT-HPNE, AC16, ARPE-19, MCF 10A, and WPMY-1 were seeded onto 96-well plates at a density of 10⁴ cells/well for 24 h and then separately cultivated with various concentrations (0–30 µg·mL^–1) of different drugs (apremilast, hydrochlorothiazide, progesterone, dexamethasone, ibuprofen, colchicine, rosiglitazone). The percentage of DMSO did not exceed 0.20% in the medium. After an additional 48 h incubation, the medium was removed, and the cells were further treated with the CCK-8 solution for another 1.5 h. In the end, a microplate reader was used to detect cell viability.

Antitumor activity in vitro

Briefly, A549, SW-1573, NCL-H1993, NCL-H2087, NCL-H3122, NCL-H1650, NCL-H1755, NCL-H1944, NCL-H1975, NCL-H23, and NCL-H358 cells were incubated in 96-well plates at a density of 10⁴ cells/well for 24 h. Afterward, various concentrations of (0–30 µg·mL^− 1) of different drugs (apremilast, hydrochlorothiazide, progesterone, dexamethasone, ibuprofen, colchicine, rosiglitazone) were incubated with the cells for 48 h. Finally, the cell viability was tested with the CCK-8 array after treatment with the CCK-8 solution for another 1.5 h.

The cell survival rate is calculated as follows:

As: Absorbance of the experimental hole (culture medium containing cells, CCK-8, DMSO, drug to be tested)

Ac: Absorbance of the control hole (medium containing cells, CCK-8, DMSO, no drug to be tested)

Ab: Absorbance of blank hole (culture medium, CCK-8, DMSO, without cells and drug to be tested)

Using a three-step strategy to prioritize cancer-type specific genes and identify potential drugs

We proposed the first systematic investigation of repurposed drugs targeting genes, specifically essential in 19 different cancer types. The general framework of this study is outlined in Fig. 1. First, we obtained information on whether the 7470 genes were essential in each of the 325 cell lines, reported by the Sanger [22] Center and excluded 13 cancer types with cell line numbers less than five in that gene dependence project. Then, we counted the EssentialitySpecificityScore (e score) of these genes in each cancer type. With this score, we constructed relationships between genes and cancer types. We also considered the ratio of cell lines for which a gene is essential in a cancer type and defined it as n in the formula of scoring function for the e score. With the two scores, we will retain only those genes larger than certain thresholds as cancer-type specific targets.

The methodology for obtaining gene-drug records refers to the PanDrugs (http://www.Pandrugs.org) [7]. Gene-drug records indicate whether a gene is a potential target of a drug. The specific source of gene-drug records and the number of records in each source are shown in Table S1. For all the cancer-type specific gene targets, we match them with drugs based on gene-drug records.

Meanwhile, referring to SLKG [23], DrugScore was calculated as described in the methods to integrate the drug research and marketing (R&M) status and target gene type, and the scoring criteria for DrugScore are shown in Table 1. When matching drugs to specific cancers through gene-drug records, drugs with DrugScore less than 0.8 were filtered out.

Hence, we identified drugs for each cancer type based on the relationship between genes and cancer types, gene-drug records, as well as the R&M status of drugs. All the identified drugs have n > 0.5 to guarantee the wide-spectrum in a given cancer type; e > 1.3 means the high specificity and less toxicity associated with inhibiting the target gene; DrugScore ≥ 0.8 reflects the clinical safety. Among the 19 cancer types, the numbers of identified drugs and their corresponding targets are shown in Table S2. All information was uploaded onto the EGKG (Essential Gene Knowledge Graph) (http://guolab.whu.edu.cn/egkg/index.php), which presents a potential option for precision treatment of tumors based on the specificity of essential genes. For example, there are a total of 39 lung adenocarcinoma-specific targets; among them, seven have matching drugs. However, only five have matching drugs with DrugScore larger than or equal to 0.8, and the number of so-called lung adenocarcinoma-specific drugs is 32.

Among our chosen 1151 cancer-type specific targets, most have no matched drugs. We also listed these 1051 targets without bound drugs in 19 cancer types. For example, gene YPEL5 has a n value of 0.8 and an e score of 1.656 for lung adenocarcinoma. If, in the future, novel drugs could be developed targeting it, they would be wide-spectrum in lung adenocarcinoma while may cause less toxicity because of the high specificity of essentiality for this gene in the cancer type. Novel drugs targeting Gene SEC63 (n: 0.650, e: 2.401) are also expected to treat lung adenocarcinoma with high safety.

The workflow of screening cancer type-specific targets and identifying drug candidates by gene specificity scores, gene-drug interaction information and drug scores is shown in Fig. 2A and Fig. 2B, respectively. Consequently, among the finally identified targets and drugs, there are different numbers of their applied cancer types. For example, 368 identified targets are applicable to only one cancer type and all identified targets (both with or without matching drugs, Fig. 2C and Fig. 2D) are applied to less than 10 cancer types, indicating these targets’ specificity and agents specifically targeting them would cause lower toxicity. As shown in Fig. 2E, there are 69 identified drugs applied to only one cancer type and the cumulative number of identified drugs applicable to nine or fewer cancer types is 335, which accounts for 90.5% of all identified drugs. Although some drugs may bind multiple essential genes and we consider their congregate effects in our screening procedure, rare wide-spectrum anti-cancer drugs appear in our final list. In summary, it can be observed that the targets we identified have prominent cancer-type specificity, and the drugs identified also have visible cancer-type specificity, which is consistent with our theoretical speculation.

Table 2

The sources of gene-drug records, e score, *DrugScore* of seven chosen drugs to validate, and their affinities with interacting targets in lung adenocarcinoma.
Drugs	Target	Source	Source provider	n score	e score	DrugScore	AA (kcal/mol)
Apremilast	CDK4	TTD	TTD	0.55	2.13	1	-8.2
Colchicine	SCD	NCI	DGIdb	0.8	1.32	1	-6.2
Dexamethasone	CDK4	NCI, CIViC	DGIdb	0.55	2.13	1	-9.2
Hydrochlorothiazide	WNK1	PharmGKB	DGIdb	0.5	1.56	1	-6.1
Ibuprofen	VHL	NCI	DGIdb	0.65	1.33	1	-5.1
Progesterone	CDK4	NCI	DGIdb	0.55	2.13	1	-9.9
Rosiglitazone	SCD	NCI	DGIdb	0.8	1.32	1	-8.9
Note: AA: average affinity (kcal/mol); ABS: absolute value

Identified drugs were validated with in vitro pharmacological evidence

For lung adenocarcinoma, we selected the top six genes (CDK4, WNK1, HSP90B1, OGDH, VHL, SCD) with the highest e-value to match with interacting drugs (among the first six genes, there were no targeting drugs for HSP90B1 and OGDH). After removing the existing cancer applications (Table S3) (TTD) as well as the drugs that were not available, we finally selected seven drugs out of the top 32 drugs with the highest e value for cell experiments and listed their information in Table 2. In addition, CDK4 and SCD are basically complementarily essential in 11 lung adenocarcinoma cell lines (Table S4), which theoretically have the potential to be combined. In other words, the two genes have orthogonal essentiality profiles across lung cancer cell lines. So, their targeting drugs, progesterone and rosiglitazone, were combined to treat lung adenocarcinoma cell lines.

In fact, top-identified drugs often have been validated through clinical trial results on the ClinicalTrials.gov website. Of the top 32 reusable lung adenocarcinoma drugs, 23 are in clinical trials, with 21 having cancer clinical trials. Among the top 100 best reusable lung adenocarcinoma drugs, 86 were involved in clinical trials, with 76 having cancer clinical trials. These results suggest that a significant proportion of the top-ranked reusable drugs are registered in clinical trials, further demonstrating the reliability of our predictive results (Table S5).

In order to validate the virtual screening results of EGKG for lung adenocarcinoma, the top seven drugs were selected to be experimentally validated in vitro. Eleven lung adenocarcinoma cell lines are involved to test drug efficacy. In clinical practice, when drugs target cancer essential genes, they may cause a series of side effects and toxic reactions, such as lipid metabolism disorders and liver damage, because the target gene may play vital functions also for normal cells. To address the above problems, in this study, we also verified the effect of these seven drugs on ten normal cell lines from different parts of the human body.

In the anti-lung adenocarcinoma experiment, the average survival rate (Fig. 3A) and standard deviation of 11 lung adenocarcinoma cell lines and 10 normal cell lines, and the p-value of the Student’s t-test for two types of cells were shown in Table S6. The survival rate of each lung adenocarcinoma cell line and normal cell line is also shown (Figs. 3C, 3D, 3E, 3F, figs. S1 and S2). It can be observed that after treatment with seven drugs, the cell survival rate of lung cancer cell lines is significantly reduced compared with normal cell lines. More specifically, colchicine has significant anti-cancer efficacy with no significant toxic side effects on normal cells (Fig. 3C, and Fig. 3D). In addition, p value of the Student’s t-test of colchicine survival rate was 0.00066 (Table S6), indicating that its effect on cancer cell lines and normal cell lines was significantly different. We also used IC50 values as parameters for drug efficacy, and the results of IC50 values for these drugs are shown in Table S7. Colchicine can be seen to have lower IC50 values for 11 lung cancer cell lines compared to the other six drugs (Fig. 3B). From the above results, it can be seen that colchicine has an excellent killing effect on a variety of lung cancer cell lines and has good biocompatibility with normal cell lines.

Because their targeting genes CDK4 and SCD have orthogonal essentialities across cell lines of lung adenocarcinoma, it is important to test the combination effect of progesterone and rosiglitazone. Then, we verified this in vitro on 11 lung adenocarcinoma cell lines. Clinically, rosiglitazone is indicated for diabetes mellitus. In the combination drug experiment, the average inhibition rate of 11 lung adenocarcinoma cell lines could reach 68.77% at a concentration of 30 µg·mL^− 1. When these two drugs were used alone, the average inhibition rates of progesterone and rosiglitazone against lung cancer cell lines were 53.64% and 47.85%, respectively (Table S6). It can be seen that the combination of drugs has a more significant anti-cancer effect than the use of drugs alone (Fig. 3E, fig. S1C, and fig. S2C). Subsequently, we further verified the effect of the combination of progesterone and rosiglitazone on the viability of a variety of normal cell lines (Fig. 3F), and the average cell survival rate was 94.31% (Table S6). In addition, the p-value of the Student’s t-test of the survival rate of combined treatment with progesterone and rosiglitazone was 1.26E-11 (Table S6), indicating that there was a significant difference in its effect on cancer cell lines and normal cell lines. The above results indicate that the drug combination has a stronger anti-cancer effect than the use alone, and the biocompatibility is superior.

Signal pathway associated with CDK4 and SCD and use of them as targets

CDK4 (Cyclin-dependent kinase 4) is a cell cycle-regulated protein kinase that binds to cyclin D and promotes cell entry into the S phase and initiates DNA synthesis during the G1 phase of the cell. In addition to regulating the cell cycle, CDK4 is also associated with PI3K/Akt, Ras/Raf/MAPK, and other signaling pathways, affecting cell proliferation and growth. Aberrant CDK4 activity is associated with tumorigenesis and progression [35]. Therefore, CDK4 is considered to be a potential target for cancer treatment, and inhibitors targeting CDK4 have become one of the research hotspots in tumor therapy.

SCD (stearoyl-CoA desaturase) is a fatty acid desaturase enzyme responsible for converting saturated fatty acids into monounsaturated fatty acids. This conversion is one of the key steps in the synthesis of triglycerides and phospholipids. The activity of SCD is controlled by a variety of factors, including nutritional status and insulin levels. Aberrant SCD activity may be associated with obesity, diabetes, and certain cancers [36–38]. As a result, SCD has emerged as a potential drug target, and inhibitors targeting it have been studied for the treatment of related diseases. The molecular docking sites of SCD mainly include the active center, substrate binding site, and other sites that may interact with regulatory factors. The substrate binding site is one of the important regions of molecular docking. Molecular docking and molecular dynamic simulation were carried out based on the original ligand binding sites on the crystal structures of CDK4 and SCD.

Molecular docking and molecular dynamic (MD) simulation of SCD with colchicine and rosiglitazone, CDK4 with progesterone

Following the above screening procedure, we obtained a total of seven drug candidates targeting lung adenocarcinoma cell lines. We obtained preliminary affinity calculations by molecular docking (Table 2). Previous studies [39] suggest that the mechanism by which gemcitabine improves lung cancer may be related to targeting the RRM1 protein (ribonucleotide reductase M1). However, the mechanism of action of our chosen drugs on lung cancer cell lines is unclear. Based on gene-drug interactions deposited in the public database, we picked out SCD for colchicine and rosiglitazone, and CDK4 for progesterone. SCD and CDK4, which are widely essential in lung cancer cell lines, were nine and seven in 11 lung cancer cell lines, respectively (Table S4). To explore the role of the three drugs with their potential targets in lung cancer cell lines, we performed molecular docking and MD simulation experiments sequentially. The results of molecular docking showed that the binding energy of colchicine to SCD protein was − 6.2 kcal/mol, the binding energy of progesterone to CDK4 protein was − 9.9 kcal/mol, and the binding energy of rosiglitazone to SCD protein was − 8.9 kcal/mol, indicating that the three drugs could bind well to their corresponding targets.

In addition, molecular docking models of progesterone with CDK4 (PDB ID: 7SJ3) and colchicine and rosiglitazone with SCD (PDB ID: 4ZYO), respectively, showed that colchicine, rosiglitazone, and progesterone docked to the binding site of the SCD original ligand and the binding site of the CDK4 original ligand, respectively (Fig. 4). This finding confirms that SCD may be a key target for colchicine to kill lung cancer cells and also confirms that CDK4 and SCD are possible targets for the combination of progesterone with rosiglitazone to kill lung cancer cells.

Root mean square deviation (RMSD), radius of gyration (Rg), total energy, and solvent accessible surface area (SASA) reflect the overall stability of the drug in the protein pocket and the conformational changes of the protein itself. The molecular docking results showed (Fig. 4) that colchicine, progesterone, and rosiglitazone could bind to the original ligand binding sites for SCD, CDK4, and SCD, respectively. In this regard, our MD results support the conclusion of molecular docking. In the MD simulation results (fig. S3, S4, and S5), the three drugs were moved in the corresponding protein cavities with little change in total energy (fig. S3C, S4C, and S5C). However, in the RMSD (fig. S3A, S4A, and S5A), Rg (fig. S3B, S4B, and S5B), and SASA results (fig. S3D, S4D, and S5D), the three drugs are moved in the corresponding proteins, and the structure of the proteins can remain in a stable state. In summary, the MD simulation results showed that the original ligand site stability of the three drugs corresponding to their targets was high.

In addition, the RMSD/Cluster index plot during the 50 ns simulation and the protein number density map both suggest that the protein conformations can remain stable after binding the three drugs at their original ligand binding sites (fig. S6). As a result, the probability of binding to the original ligand sites corresponding to their targets is very high, and the interpretation of this conclusion combines affinity and MD simulation results.

Most of the 370 identified drugs are repurposed as anti-cancer agents and have a restricted number of bound targets

We have identified 370 cancer-type specific essential genes bound drugs, which are distributed among 19 cancer types. Some of these drugs’ originally approved indications may be cancer therapy, while others may not. We extracted the information on treating indications from Drugbank and showed their distribution in Fig. 5A. As we can see, only 24.3% of drugs are developed for tumor therapy and over 70% are identified as anti-tumor because they may bind targets that are essential in specific cancer cells (types). These results could preliminarily demonstrate that our identifying strategy has the capacity of drug repurposing.

Ideal drugs are expected to have specific and precise targeting, while off-targeting would cause toxicity to normal cells or the human body [40]. This rule holds valid for our identified anti-cancer drugs. Since we assess the specificity of one drug by each of all potentially bound targets, if this drug could bind simultaneously with many essential genes, its specificity would be compromised. In Fig. 5B, we show the distribution of the number of recorded targets that are essential in any cancer cells for all 370 drugs. It is acceptable that about half of the 370 drugs have essential targets in one to three, indicating relatively high specificity of these drugs as anti-tumor agents.

We think that drug toxicity to normal cells is determined by the aggregative essentiality of all simultaneously bound targets in all cancer cells. In the inhibitory experiments, we observed a correlation (Pearson's correlation analysis, R = -0.224, p = 0.062, 77 inhibitory rates with 7 aggregative m values of all validated drugs) between the aggregative m values of potentially bound targets and the viability of the normal cells. We think that the observed correlation is not highly significant just because the tested drug number is as low as seven. Based on this result if a drug has m values less than 0.5, it could not hold high inhibitory efficiency on normal cells. Therefore, if we figure out the aggregative essentiality of all the potential targets for one drug, we can roughly estimate its maximum toxicity to normal cells. Hence, and we sum the essential number of all potential targets recorded in the public databases for all drugs. For example, prednisone has 10 binding targets and the m value of the aggregative 10 targets is 0.41 (132/325). Consequently, we got Fig. 5C and luckily found that about half of 370 drugs have maximum essentiality (m value) less than 0.5. In fact, a drug could not simultaneously bind all potential targets, and hence, much more drugs will be safe because their factual aggregative essentiality will be less than 0.5.

Many studies have shown that essential genes can be used for antimicrobial drug targets [21], but no method has been proposed to screen anticancer drugs based on a single essential gene. To this end, we have developed EGKG, which would help the precision treatment of cancers based on essential genes from the perspective of drug repurposing and provides a new idea for developing anticancer drugs targeting cancer-type specific essential genes. We validated the effectiveness of all the seven drugs identified by this strategy with lung adenocarcinoma cell inhibiting experiments. More importantly, it was found that there are significantly different (p = 0.0001) inhibiting rates between cancer cells and normal cells for the seven drugs. This suggest that those identified drugs with higher e values could have higher inhibiting rates on associated cancer cells than on normal cells and could be regarded as having high inhibiting specificity or treatment safety. Furthermore, we observed a weakly negative correlation (p = 0.062) for the aggregative targets between the aggregative m values of potentially bound targets and the viability of normal cells. Hence, the m value of one essential gene could partly reflect the safety of the drug when binding to the target gene, and the e score helped to identify those drugs having lower inhibiting rate on associated cancer cells than normal cells. These results validated our supposition that cancer-type specific essential genes could be regarded as ideal targets, whose bound drugs have less toxicity than those bound to less specific essential genes. In fact, precision medicine needs more anti-tumor drugs with higher safety and lower body toxicity [41]. Although it has been implied that drugs targeting specific essential genes could meet this requirement, here for the first time, we performed a systematic investigation of ranking drugs targeting cancer-type specific essential genes and identified hundreds of such drugs applied in 19 cancer types.

Combinational therapy by multiple drugs is an emerging intervention and several drugs may play roles through different mechanisms [42]. Here, we validated a novel combining strategy by targeting simultaneously genes with orthogonal essentialities. Consequently, treating the lung adenocarcinoma cell lines and normal cells simultaneously with both progesterone and rosiglitazone, the inhibiting rate for the former increased whereas that for the latter decreased. Hence, this way of targeting genes with orthogonal essentialities could improve chemotherapy and reduce damage to normal cells.

These findings still need to be further updated and improved: (1) Besides proof from literature, large-scale in vitro and/or in vivo experiments are expected to validate our identified 370 cancer-type specific drugs. (2) Find more examples that multiple targets with orthogonal essentialities can be used in combination with drugs and conduct experiments to verify their efficacy. (3) Among the 19 cancer types, there are still 1051 targets without bound drugs. These cancer-type specific targets are listed on the EGKG site and this information, particularly on those targets with top rank, deserves special attention from researchers in the fields of targeted therapy and vaccine design.

We proposed a strategy to systematically identify cancer-type specific genes and drugs targeting these genes. Among 19 cancer types, we identified 1151 cancer-type specific targets and 370 drugs bound with 100 of these targets. Profiles of applied cancer types of identified targets and drugs illustrate the effectiveness of the scoring strategy and most of them apply to limited cancer types. Our cell experiment results suggest that cancer-type specific essential genes are ideal targets, and when having a similar effect on tumors, drugs definitely bound to these genes would constitute the preferred option of precision treatment because of less cellular toxicity. Our identified target and drug lists are freely accessed at the EGKG site (http://guolab.whu.edu.cn/egkg/index.php) and the EGKG forms a computational basis to uncover essential gene targets and drugs applicable to specific cancer types.

Acknowledgments

We thank Miss Candy Sizhe Wu at Cornell University for polishing the language of our manuscript.

Funding

This work was supported by the National Natural Science Foundation of China (grant

no. 32370696).

Author contributions statement

F.G. conceived the concept and supervised this project. X.L. designed the cell experiments and performed them under the supervision of Y.L., Z.D., and Z.C.. X.K. collected data, D.Z., Q.X., A.Y., C.W., H.C., and H.G. helped with data collection. X.L. and F.G. wrote the paper with input from other authors. J.Z. maintained the server.

Competing interests

The authors declare that they have no other competing interests.

Data and materials availability

All data associated with this study are present in the paper or Supplementary Materials. The code used to process the data in EGKG of this work has been uploaded in FigShare (https://figshare.com/articles/software/EGKG_data_process/26771053?file=48632788).

Hay M, Thomas DW, Craighead JL, Economides C, Rosenthal J. Clinical development success rates for investigational drugs. Nat. Biotechnol. 2014;32(1):40-51.
Kong J, Lee H, Kim D, Han SK, Ha D, Shin K, Kim S. Network-based machine learning in colorectal and bladder organoid models predicts anti-cancer drug efficacy in patients. Nat Commun. 2020;11(1):5485.
Zhou Z, Li M. Targeted therapies for cancer. BMC Med. 2022;20(1):90.
Dai M, Yan G, Wang N, Daliah G, Edick AM, Poulet S, Boudreault J, Ali S, Burgos SA, Lebrun JJ. In vivo genome-wide CRISPR screen reveals breast cancer vulnerabilities and synergistic mTOR/Hippo targeted combination therapy. Nat Commun. 2021;12(1):3055.
Schenone M, Dančík V, Wagner BK, Clemons PA. Target identification and mechanism of action in chemical biology and drug discovery. Nat Chem Biol. 2013;9(4):232-40.
Konda P, Garinet S, Van Allen EM, Viswanathan SR. Genome-guided discovery of cancer therapeutic targets. Cell Rep. 2023;42(8):112978.
Piñeiro-Yáñez E, Reboiro-Jato M, Gómez-López G, Perales-Patón J, Troulé K, Rodríguez JM, Tejero H, Shimamura T, López-Casas PP, Carretero J, Valencia A, Hidalgo M, Glez-Peña D, Al-Shahrour F. PanDrugs: a novel method to prioritize anticancer drug treatments according to individual genomic data. Genome Med. 2018;10(1):41.
Kuijk E, Kranenburg O, Cuppen E, Van Hoeck A. Common anti-cancer therapies induce somatic mutations in stem cells of healthy tissue. Nat Commun. 2022;13(1):5915.
Zehir A, Benayed R, Shah RH, Syed A, Middha S, Kim HR, Srinivasan P, Gao J, Chakravarty D, Devlin SM, Hellmann MD, Barron DA, Schram AM, Hameed M, Dogan S, Ross DS, Hechtman JF, DeLair DF, Yao J, Mandelker DL, Cheng DT, Chandramohan R, Mohanty AS, Ptashkin RN, Jayakumaran G, Prasad M, Syed MH, Rema AB, Liu ZY, Nafa K, Borsu L, Sadowska J, Casanova J, Bacares R, Kiecka IJ, Razumova A, Son JB, Stewart L, Baldi T, Mullaney KA, Al-Ahmadie H, Vakiani E, Abeshouse AA, Penson AV, Jonsson P, Camacho N, Chang MT, Won HH, Gross BE, Kundra R, Heins ZJ, Chen HW, Phillips S, Zhang H, Wang J, Ochoa A, Wills J, Eubank M, Thomas SB, Gardos SM, Reales DN, Galle J, Durany R, Cambria R, Abida W, Cercek A, Feldman DR, Gounder MM, Hakimi AA, Harding JJ, Iyer G, Janjigian YY, Jordan EJ, Kelly CM, Lowery MA, Morris LGT, Omuro AM, Raj N, Razavi P, Shoushtari AN, Shukla N, Soumerai TE, Varghese AM, Yaeger R, Coleman J, Bochner B, Riely GJ, Saltz LB, Scher HI, Sabbatini PJ, Robson ME, Klimstra DS, Taylor BS, Baselga J, Schultz N, Hyman DM, Arcila ME, Solit DB, Ladanyi M, Berger MF. Mutational landscape of metastatic cancer revealed from prospective clinical sequencing of 10,000 patients. Nat. Med. 2017;23(6):703-713.
Mirahmadi M, Azimi-Hashemi S, Saburi E, Kamali H, Pishbin M, Hadizadeh F. Potential inhibitory effect of lycopene on prostate cancer. Biomed. Pharmacother. 2020;129:110459.
Coombes RC. Drug testing in the patient: toward personalized cancer treatment. Sci Transl Med. 2015;7(284):284ps10.
Patel SJ, Sanjana NE, Kishton RJ, Eidizadeh A, Vodnala SK, Cam M, Gartner JJ, Jia L, Steinberg SM, Yamamoto TN, Merchant AS, Mehta GU, Chichura A, Shalem O, Tran E, Eil R, Sukumar M, Guijarro EP, Day CP, Robbins P, Feldman S, Merlino G, Zhang F, Restifo NP. Identification of essential genes for cancer immunotherapy. Nature. 2017;548(7669):537-542.
Funk L, Su KC, Ly J, Feldman D, Singh A, Moodie B, Blainey PC, Cheeseman IM. The phenotypic landscape of essential human genes. Cell. 2022;185(24):4634-4653.e22.
Juhas M, Eberl L, Church GM. Essential genes as antimicrobial targets and cornerstones of synthetic biology. Trends Biotechnol. 2012;30(11):601-7.
Ji X, Rajpal DK, Freudenberg JM. The essentiality of drug targets: an analysis of current literature and genomic databases. Drug Discov Today. 2019;24(2):544-550.
Tyner JW. Functional genomics for personalized cancer therapy. Sci Transl Med. 2014;6(243):243fs26.
Nguyen TTT, Kim YT, Jeong G, Jin M. Immunopathology of and potential therapeutics for secondary hemophagocytic lymphohistiocytosis/macrophage activation syndrome: a translational perspective. Exp Mol Med. 2024;56(3):559-569.
Gonzalez-Fierro A, Dueñas-González A. Drug repurposing for cancer therapy, easier said than done. Semin Cancer Biol. 2021;68:123-131.
Wu Y, Zhang C, Liu X, He Z, Shan B, Zeng Q, Zhao Q, Zhu H, Liao H, Cen X, Xu X, Zhang M, Hou T, Wang Z, Yan H, Yang S, Sun Y, Chen Y, Wu R, Xie T, Chen W, Najafov A, Ying S, Xia H. ARIH1 signaling promotes anti-tumor immunity by targeting PD-L1 for proteasomal degradation. Nat Commun. 2021;12(1):2346.
Mayer RL, Verbeke R, Asselman C, Aernout I, Gul A, Eggermont D, Boucher K, Thery F, Maia TM, Demol H, Gabriels R, Martens L, Bécavin C, De Smedt SC, Vandekerckhove B, Lentacker I, Impens F. Immunopeptidomics-based design of mRNA vaccine formulations against Listeria monocytogenes. Nat Commun. 2022;13(1):6075.
Wang R, Shu X, Zhao H, Xue Q, Liu C, Wu A, Cheng F, Wang L, Zhang Y, Feng J, Wu N, Li M. Associate toxin-antitoxin with CRISPR-Cas to kill multidrug-resistant pathogens. Nat Commun. 2023;14(1):2078.
Behan FM, Iorio F, Picco G, Gonçalves E, Beaver CM, Migliardi G, Santos R, Rao Y, Sassi F, Pinnelli M, Ansari R, Harper S, Jackson DA, McRae R, Pooley R, Wilkinson P, van der Meer D, Dow D, Buser-Doepner C, Bertotti A, Trusolino L, Stronach EA, Saez-Rodriguez J, Yusa K, Garnett MJ. Prioritization of cancer therapeutic targets using CRISPR-Cas9 screens. Nature. 2019;568(7753):511-516.
Zhang B, Tang C, Yao Y, Chen X, Zhou C, Wei Z, Xing F, Chen L, Cai X, Zhang Z, Sun S, Liu Q. The tumor therapy landscape of synthetic lethality. Nat Commun. 2021;12(1):1275.
Chang L, Ruiz P, Ito T, Sellers WR. Targeting pan-essential genes in cancer: Challenges and opportunities. Cancer Cell. 2021;39(4):466-479.
Setton J, Zinda M, Riaz N, Durocher D, Zimmermann M, Koehler M, Reis-Filho JS, Powell SN. Synthetic lethality in cancer therapeutics: the next generation. Cancer Discov. 2021;11(7):1626-1635.
Martínez-Jiménez F, Muiños F, Sentís I, Deu-Pons J, Reyes-Salazar I, Arnedo-Pac C, Mularoni L, Pich O, Bonet J, Kranas H, Gonzalez-Perez A, Lopez-Bigas N. A compendium of mutational cancer driver genes. Nat Rev Cancer. 2020;20(10):555-572.
Bates SE. Epigenetic Therapies for Cancer. N Engl J Med. 2020;383(7):650-663.
Terekhanova NV, Karpova A, Liang WW, Strzalkowski A, Chen S, Li Y, Southard-Smith AN, Iglesia MD, Wendl MC, Jayasinghe RG, Liu J, Song Y, Cao S, Houston A, Liu X, Wyczalkowski MA, Lu RJ, Caravan W, Shinkle A, Naser Al Deen N, Herndon JM, Mudd J, Ma C, Sarkar H, Sato K, Ibrahim OM, Mo CK, Chasnoff SE, Porta-Pardo E, Held JM, Pachynski R, Schwarz JK, Gillanders WE, Kim AH, Vij R, DiPersio JF, Puram SV, Chheda MG, Fuh KC, DeNardo DG, Fields RC, Chen F, Raphael BJ, Ding L. Epigenetic regulation during cancer transitions across 11 tumour types. Nature. 2023;623(7986):432-441.
Hahn WC, Bader JS, Braun TP, Califano A, Clemons PA, Druker BJ, Ewald AJ, Fu H, Jagu S, Kemp CJ, Kim W, Kuo CJ, McManus M, B Mills G, Mo X, Sahni N, Schreiber SL, Talamas JA, Tamayo P, Tyner JW, Wagner BK, Weiss WA, Gerhard DS; Cancer Target Discovery and Development Network. An expanded universe of cancer targets. Cell. 2021;184(5):1142-1155.
Dietlein F, Wang AB, Fagre C, Tang A, Besselink NJM, Cuppen E, Li C, Sunyaev SR, Neal JT, Van Allen EM. Genome-wide analysis of somatic noncoding mutation patterns in cancer. Science. 2022;376(6589):eabg5601.
Zhou Y, Zhang Y, Lian X, Li F, Wang C, Zhu F, Qiu Y, Chen Y. Therapeutic target database update 2022: facilitating drug discovery with enriched comparative data of targeted agents. Nucleic Acids Res. 2022;50(D1):1398-D1407.
Pacini C, Duncan E, Gonçalves E, Gilbert J, Bhosle S, Horswell S, Karakoc E, Lightfoot H, Curry E, Muyas F, Bouaboula M, Pedamallu CS, Cortes-Ciriano I, Behan FM, Zalmas LP, Barthorpe A, Francies H, Rowley S, Pollard J, Beltrao P, Parts L, Iorio F, Garnett MJ. A comprehensive clinically informed map of dependencies in cancer cells and framework for target prioritization. Cancer Cell. 2024;42(2):301-316.e9.
Freshour SL, Kiwala S, Cotto KC, Coffman AC, McMichael JF, Song JJ, Griffith M, Griffith OL, Wagner AH. Integration of the drug-gene interaction database (DGIdb 4.0) with open crowdsource efforts. Nucleic Acids Res. 2021;49(D1):D1144-D1151.
Wishart DS, Feunang YD, Guo AC, Lo EJ, Marcu A, Grant JR, Sajed T, Johnson D, Li C, Sayeeda Z, Assempour N, Iynkkaran I, Liu Y, Maciejewski A, Gale N, Wilson A, Chin L, Cummings R, Le D, Pon A, Knox C, Wilson M. DrugBank 5.0: a major update to the DrugBank database for 2018. Nucleic Acids Res. 2018;46(D1):D1074-D1082.
Chen P, Xu Y, Li X, Yao H, Lin K. Development and strategies of CDK4/6 inhibitors. Future Med Chem. 2020;12(2):127-145.
Sun Q, Xing X, Wang H, Wan K, Fan R, Liu C, Wang Y, Wu W, Wang Y, Wang R. SCD1 is the critical signaling hub to mediate metabolic diseases: mechanism and the development of its inhibitors. Biomed Pharmacother. 2024;170:115586.
Kim SJ, Choi H, Park SS, Chang C, Kim E. Stearoyl CoA desaturase (SCD) facilitates proliferation of prostate cancer cells through enhancement of androgen receptor transactivation. Mol Cells. 2011;31(4):371-377.
Bansal S, Berk M, Alkhouri N, Partrick DA, Fung JJ, Feldstein A. Stearoyl-CoA desaturase plays an important role in proliferation and chemoresistance in human hepatocellular carcinoma. J Surg Res. 2014;186(1):29-38.
Bepler G, Kusmartseva I, Sharma S, Gautam A, Cantor A, Sharma A, Simon G. RRM1 modulated in vitro and in vivo efficacy of gemcitabine and platinum in non-small-cell lung cancer. J Clin Oncol. 2006;24(29):4731-7.
Fang RH, Gao W, Zhang L. Targeting drugs to tumours using cell membrane-coated nanoparticles. Nat Rev Clin Oncol. 2023;20(1):33-48.
Galetin A, Brouwer KLR, Tweedie D, Yoshida K, Sjöstedt N, Aleksunes L, Chu X, Evers R, Hafey MJ, Lai Y, Matsson P, Riselli A, Shen H, Sparreboom A, Varma MVS, Yang J, Yang X, Yee SW, Zamek-Gliszczynski MJ, Zhang L, Giacomini KM. Membrane transporters in drug development and as determinants of precision medicine. Nat Rev Drug Discov. 2024;23(4):255-280.
Chen S, Xie J, Ye R, Xu DD, Yang Y. Structure-aware dual-target drug design through collaborative learning of pharmacophore combination and molecular simulation. Chem Sci. 2024;15(27):10366-10380.
Eberhardt J, Santos-Martins D, Tillack AF, Forli S. AutoDock vina 1.2.0: new docking methods, expanded force field, and python bindings. J Chem Inf Model. 2021;61(8):3891-3898.
Chen C, Chen H, Zhang Y, Thomas HR, Frank MH, He Y, Xia R. TBtools: An integrative toolkit developed for interactive analyses of big biological data. Mol Plant. 2020;13(8):1194-1202.
Zhang D, Wang Z, Li J, Zhu J. Exploring the possible molecular targeting mechanism of Saussurea involucrata in the treatment of COVID-19 based on bioinformatics and network pharmacology. Comput Biol Med. 2022;146:105549.
Lee WY, Chen KC, Chen HY, Chen CY. Potential mitochondrial isocitrate dehydrogenase R140Q mutant inhibitor from traditional Chinese medicine against cancers. Biomed Res Int. 2014;2014:364625.
Hess B, Kutzner C, van der Spoel D, Lindahl E. GROMACS 4: algorithms for highly efficient, load-balanced, and scalable molecular simulation. J Chem Theory Comput. 2008;4(3):435-47.

No competing interests reported.

SupplementaryMaterials.docx
List of Supplementary Materials The PDF file includes: Supplementary Materials and Methods Figs. S1 to S7 Tables S1 to S7 References (43-47)

Download PDF

Version 1

posted

You are reading this latest preprint version

Systematic identification of cancer-type specific drugs based on essential genes and validations in lung adenocarcinoma

Status:

Version 1

Abstract

Figures

Background

Methods and materials

Building of EGKG

Entity and relationship establishment of EGKG

Drug reorientation scoring model

EGKG web server development

Materials

Cell culture

Results

Using a three-step strategy to prioritize cancer-type specific genes and identify potential drugs

Signal pathway associated with CDK4 and SCD and use of them as targets

Discussion

Conclusions

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1