Rapid proteomic analysis for solid tumors reveals LSD1 as a drug target in an end‐stage cancer patient

Doll et al. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6068348/

Abstract

Recent advances in mass spectrometry (MS)‐based technologies are now set to transform translational cancer proteomics from an idea to a practice. Here, we present a robust proteomic workflow for the analysis of clinically relevant human cancer tissues that allows quantitation of thousands of tumor proteins in several hours of measuring time and a total turnaround of a few days. We applied it to a chemorefractory metastatic case of the extremely rare urachal carcinoma. Quantitative comparison of lung metastases and surrounding tissue revealed several significantly upregulated proteins, among them lysine‐specific histone demethylase 1 (LSD1/KDM1A). LSD1 is an epigenetic regulator and the target of active development efforts in oncology. Thus, clinical cancer proteomics can rapidly and efficiently identify actionable therapeutic options. While currently described for a single case study, we envision that it can be applied broadly to other patients in a similar condition.

Workflow

Figure 2 Proteomics workflow for the case study. (A) Timeline of the project. (B) Experimental design, including source of material, inStageTip sample preparation, and depiction of the analytical workflow

Results

Figure 3 Proteins differentially expressed in the urachal carcinoma lung metastases. (A) Volcano plot of the p‐values (y‐axis) vs. the log2 protein abundance differences (x‐axis) between metastases and control, with lines of significance colored in black or gray lines corresponding to a 5% or 1% FDR, respectively. (B) Mechanisms of action of LSD1/KDM1A and inhibitory drug treatment proposed: JATROSOME. TRANYLCYPROMIN

Workflow with the Clinical Knowledge Graph

1. Generate Analysis Report: Proteomics data

2. Identify Candidate Drug Treatments

3. Rank Candidates According to Toxicity

|Clinical\_Knowledge\_Graph|

Generate Analysis Report: Proteomics Data

Report Manager

from ckg.report_manager import project

from plotly.offline import init_notebook_mode, iplot
%matplotlib inline
init_notebook_mode(connected=True)

configuration_files = {"proteomics":"../assets/proteomics_CS.yml"}

study_case_project = project.Project(identifier="P0000002", configuration_files=configuration_files, datasets={}, knowledge=None, report={})

study_case_project.build_project(force=False)

study_case_project.generate_report()

study_case_project.show_report(environment='notebook')

defaultdict(list,
            {'PROJECT INFORMATION': [],
             'PROTEOMICS': [],
             'MULTIOMICS': [],
             'KNOWLEDGE GRAPH': []})

study_case_project.list_datasets()

dict_keys(['multiomics', 'proteomics'])

proteomics_dataset = study_case_project.get_dataset(dataset='proteomics')

proteomics_dataset.list_dataframes()

['go annotation',
 'number of modified proteins',
 'number of peptides',
 'number of proteins',
 'original',
 'pathway annotation',
 'processed',
 'protein biomarkers',
 'regulated',
 'regulation table',
 'tissue qcmarkers']

In this case, we use the regulation table to extract proteins upregulated in the metastatic tissue compare to non-cancerous tissue.

regulation_table = proteomics_dataset.get_dataframe(dataset_name='regulation table')

regulation_table.head()

	-log10 pvalue	FC	Method	T-statistics	correction	dfk	dfn	effsize	efftype	group1	...	identifier	log2FC	mean(group1)	mean(group2)	padj	pvalue	rejected	s0	std(group1)	std(group2)
0	0.487413	-1.277214	SAMR Two class paired	0.153337	permutation FDR (4.0 perm)	1	2	-0.738	hedges	CANCER	...	A1BG~P04217	-0.353	31.710	32.063	0.714951	0.325527	False	2	0.047	0.384
1	1.438878	3.610003	SAMR Two class paired	-0.924233	permutation FDR (4.0 perm)	1	2	2.913	hedges	CANCER	...	A1CF~Q9NQ94	1.852	26.563	24.711	0.052925	0.036402	False	2	0.360	0.366
2	3.001177	-1.926524	SAMR Two class paired	0.463202	permutation FDR (4.0 perm)	1	2	-18.081	hedges	CANCER	...	A2M~P01023	-0.946	34.287	35.233	0.271411	0.000997	False	2	0.033	0.026
3	1.192262	2.801113	SAMR Two class paired	-0.594485	permutation FDR (4.0 perm)	1	2	2.145	hedges	CANCER	...	AAAS~Q9NRG9	1.486	26.317	24.831	0.149537	0.064230	False	2	0.173	0.532
4	0.321607	-1.786332	SAMR Two class paired	0.346611	permutation FDR (4.0 perm)	1	2	-0.496	hedges	CANCER	...	AACS~Q86V21	-0.837	26.350	27.187	0.423692	0.476862	False	2	0.625	1.211

5 rows × 21 columns

regulation_table[regulation_table['identifier']=='HDAC1~Q13547']

	-log10 pvalue	FC	Method	T-statistics	correction	dfk	dfn	effsize	efftype	group1	...	identifier	log2FC	mean(group1)	mean(group2)	padj	pvalue	rejected	s0	std(group1)	std(group2)
1885	0.782856	10.367528	SAMR Two class paired	-1.008813	permutation FDR (4.0 perm)	1	2	1.227	hedges	CANCER	...	HDAC1~Q13547	3.374	30.666	27.292	0.036735	0.164871	True	2	0.301	2.201

1 rows × 21 columns

up_regulated_proteins = regulation_table.loc[(regulation_table.rejected) & (regulation_table.FC > 2), ['identifier']]

up_regulated_proteins.shape

(199, 1)

'HDAC1~Q13547' in up_regulated_proteins['identifier'].tolist()

True

Graph Database Connector

from ckg.graphdb_connector import query_utils, connector

driver = connector.getGraphDatabaseConnectionConfiguration()

queries = query_utils.read_knowledge_queries()

1) Filter for Regulated Proteins Associated to Lung Cancer:

selected_queries = query_utils.find_queries_involving_nodes(queries=queries, nodes=["Protein", "Disease"], print_pretty=True)

selected_queries.head()

	Name	Description	involved_nodes	involved_rels	query	example
id
Disease	associated diseases in at least two of the pro...	get relationships to diseases from a list of p...	Protein,Disease	ASSOCIATED_WITH	MATCH (project:Project)-[:STUDIES_DISEASE]-(d:...
association_disease_score	specific disease	Return the list of proteins associated to a sp...	Protein,Disease	ASSOCIATED_WITH	MATCH (protein:Protein)-[r]-(disease:Disease) ...	proteins = ['A1BG~P04217','A2M~P01023','ACACB~...

disease_query = selected_queries.loc["association_disease_score", "query"]
proteins = ['"{}"'.format(p) for p in up_regulated_proteins["identifier"].tolist()]
diseases = ['DOID:1324']
diseases = ['"{}"'.format(d) for d in diseases]
disease_query = disease_query.format(",".join(proteins),",".join(diseases), 1)

proteins_associated_lung_cancer = connector.getCursorData(driver=driver, query=disease_query, parameters={})

proteins_associated_lung_cancer.head()

	node1	node2	source	type	weight
0	SLC44A1~Q8WWI5	lung cancer	DISEASES	ASSOCIATED_WITH	1.026
1	IRF6~O14896	lung cancer	DISEASES	ASSOCIATED_WITH	1.192
2	SPINT2~O43291	lung cancer	DISEASES	ASSOCIATED_WITH	1.216
3	KDM1A~O60341	lung cancer	DISEASES	ASSOCIATED_WITH	2.405
4	CEACAM5~P06731	lung cancer	DISEASES	ASSOCIATED_WITH	3.122

proteins_associated_lung_cancer.shape

(69, 5)

2) Identify Inhibitory Drugs for those Proteins

selected_queries = query_utils.find_queries_involving_nodes(queries=queries, nodes=["Protein", "Drug"], print_pretty=True)

selected_queries.head()

	Name	Description	involved_nodes	involved_rels	query	example
id
Drug	associated drugs in at least two of the protei...	get relationships to drugs. Limit the result t...	Protein,Drug	ACTS_ON	MATCH (protein:Protein)-[r:ACTS_ON]-(drug:Drug...
association_drug_intervention_proteins	drug intervention- protein association	Return associations between a list of proteins...	Project,Protein,Clinical_variable,Drug	HAD_INTERVENTION,ACTS_ON	MATCH (project:Project)-[]-()-[:HAD_INTERVENTI...	project_id = 'P0000002'\nproteins = ['A1BG~P04...
association_drug_interaction_score	drug interaction association	Return the list of drugs associated to the lis...	Protein,Drug	ACTS_ON	MATCH (protein:Protein)-[r]-(drug:Drug) WHERE ...	proteins = ['A1BG~P04217','A2M~P01023','ACACB~...

proteins = ['"{}"'.format(p) for p in proteins_associated_lung_cancer['node1'].tolist()]
drug_query = selected_queries.loc["association_drug_interaction_score", "query"].format(",".join(proteins), 'inhibition', 0.8)

drugs_proposed = connector.getCursorData(driver=driver, query=drug_query, parameters={})

drugs_proposed.head()

	Drug_desc	action	drug_id	node1	node2	source	type	weight
0	None	inhibition	DB04808	ANG~P03950	Neamine	STITCH	ACTS_ON	0.800
1	None	inhibition	DB02198	ANG~P03950	2-Bromoacetyl Group	STITCH	ACTS_ON	0.800
2	Losartan is an angiotensin II receptor blocker...	inhibition	DB00678	ANG~P03950	Losartan	STITCH	ACTS_ON	0.957
3	Tamoxifen is a non-steroidal antiestrogen used...	inhibition	DB00675	ANG~P03950	Tamoxifen	STITCH	ACTS_ON	0.990
4	A macrolide compound obtained from Streptomyce...	inhibition	DB00877	ANG~P03950	Sirolimus	STITCH	ACTS_ON	0.800

drugs_proposed.shape

(67, 8)

We can already see that CKG found the same inhibitory drug that was identified in the study case published. However, many other options are proposed and could be further ranked using other criteria.

from ckg.analytics_core import utils
from ckg.analytics_core.viz import viz

net = viz.get_network(data=drugs_proposed, identifier="inhibition_drugs", args={"source":"node1", "target":"node2", "values":"weight", "node_size":"degree","title":"Proposed drugs", "color_weight":False})

viz.visualize_notebook_network(net["notebook"], notebook_type='jupyter', layout={'width':'100%', 'height':'700px'})

utils.json_network_to_gml(net['net_json'], path='drug_network.gml')

proteins = ['"{}"'.format(p) for p in proteins_associated_lung_cancer['node1'].tolist()]
drug_query = queries["association_drug_interaction_score"]["query"].format(",".join(proteins), 'inhibition', 0.9)

drugs_proposed = connector.getCursorData(driver=driver, query=drug_query, parameters={})

drugs_proposed.head()

	Drug_desc	action	drug_id	node1	node2	source	type	weight
0	Losartan is an angiotensin II receptor blocker...	inhibition	DB00678	ANG~P03950	Losartan	STITCH	ACTS_ON	0.957
1	Tamoxifen is a non-steroidal antiestrogen used...	inhibition	DB00675	ANG~P03950	Tamoxifen	STITCH	ACTS_ON	0.990
2	Paclitaxel is a chemotherapeutic agent markete...	inhibition	DB01229	CDH1~P12830	Paclitaxel	STITCH	ACTS_ON	0.957
3	A major primary bile acid produced in the live...	inhibition	DB02659	CDH1~P12830	Cholic Acid	STITCH	ACTS_ON	0.957
4	Calcitriol is an active metabolite of vitamin ...	inhibition	DB00136	CDH17~Q12864	Calcitriol	STITCH	ACTS_ON	0.957

drugs_proposed.shape

(15, 8)

net = viz.get_network(data=drugs_proposed, identifier="inhibition_drugs", args={"source":"node1", "target":"node2", "values":"weight", "node_size":"degree","title":"Proposed drugs", "color_weight":False})

viz.visualize_notebook_network(net["notebook"], notebook_type='jupyter', layout={'width':'100%', 'height':'700px'})

utils.json_network_to_gml(net['net_json'], path='drug_network_reduced.gml')

3) Identify Proposed Drug’s Known Side Effects

selected_queries = query_utils.find_queries_involving_nodes(queries=queries, nodes=["Phenotype", "Drug"], print_pretty=True)

selected_queries.head()

	Name	Description	involved_nodes	involved_rels	query	example
id
association_drug_sideeffects	drug side effect association	Return the list of side effects linked to drugs	Phenotype,Drug	ASSOCIATED_WITH	MATCH (sideeffect:Phenotype)-[r]-(drug:Drug) W...	drugs = ['DB00439', 'DB06196']\ndrug_side_effe...

drugs = drugs_proposed["drug_id"].unique()
drugs = ['"{}"'.format(d) for d in drugs]
sideeffects_query = selected_queries.loc["association_drug_sideeffects", "query"].format(",".join(drugs))

side_effects = connector.getCursorData(driver=driver, query=sideeffects_query, parameters={})

side_effects.head()

	node1	node2	source	type
0	Losartan	Abnormality of fluid regulation	SIDER	HAS_SIDE_EFFECT
1	Losartan	Thrombocytopenia	SIDER	HAS_SIDE_EFFECT
2	Losartan	Nausea	SIDER	HAS_SIDE_EFFECT
3	Losartan	Palpitations	SIDER	HAS_SIDE_EFFECT
4	Losartan	Arthritis	SIDER	HAS_SIDE_EFFECT

side_effects.groupby('node1')['node2'].count()

node1
Atorvastatin    152
Bleomycin        51
Calcitriol       59
Cholic Acid       7
Gemcitabine      94
Glyburide        63
Losartan        120
Paclitaxel      218
Tamoxifen       106
Tolbutamide      18
Vildagliptin     22
Vorinostat       42
Name: node2, dtype: int64

net = viz.get_network(data=side_effects, identifier="side_effects", args={"source":"node1", "target":"node2", "node_size":"degree","title":"Proposed drugs", "color_weight":False})

side_effects.head()

	node1	node2	source	type	width
0	Losartan	Abnormality of fluid regulation	SIDER	HAS_SIDE_EFFECT	1
1	Losartan	Thrombocytopenia	SIDER	HAS_SIDE_EFFECT	1
2	Losartan	Nausea	SIDER	HAS_SIDE_EFFECT	1
3	Losartan	Palpitations	SIDER	HAS_SIDE_EFFECT	1
4	Losartan	Arthritis	SIDER	HAS_SIDE_EFFECT	1

utils.json_network_to_gml(net['net_json'], path='side_effects.gml')

no_registerd_side_effects = list(set(drugs_proposed['node2'].tolist()).difference(side_effects['node1'].tolist()))

no_registerd_side_effects

['Trichostatin A', 'Resveratrol', 'dATP']

4) Reduce Adverse Response

treatment_regimens = pd.DataFrame(['Oxaliplatin',
                     'Capecitabine',
                     'Folinic acid',
                     'Fluorouracil'], columns=['treatment'])

selected_queries = query_utils.find_queries_involving_nodes(queries=queries, nodes=["Clinical_variable", "Drug"], print_pretty=True)

selected_queries.head()

	Name	Description	involved_nodes	involved_rels	query	example
id
association_drug_intervention_proteins	drug intervention- protein association	Return associations between a list of proteins...	Project,Protein,Clinical_variable,Drug	HAD_INTERVENTION,ACTS_ON	MATCH (project:Project)-[]-()-[:HAD_INTERVENTI...	project_id = 'P0000002'\nproteins = ['A1BG~P04...
side_effects_jaccard_similarity_intervention_proposed_drugs	similarity between side effects	Return the jaccard similarity between drugs us...	Drug,Clinical_variable	HAS_SIDE_EFFECT	MATCH (d1:Drug)-[:HAS_SIDE_EFFECT]->(phenotype...	intervention = ['Capecitabine', 'Fluorouracil'...

drugs_proposed

	Drug_desc	action	drug_id	node1	node2	source	type	weight
0	Losartan is an angiotensin II receptor blocker...	inhibition	DB00678	ANG~P03950	Losartan	STITCH	ACTS_ON	0.957
1	Tamoxifen is a non-steroidal antiestrogen used...	inhibition	DB00675	ANG~P03950	Tamoxifen	STITCH	ACTS_ON	0.990
2	Paclitaxel is a chemotherapeutic agent markete...	inhibition	DB01229	CDH1~P12830	Paclitaxel	STITCH	ACTS_ON	0.957
3	A major primary bile acid produced in the live...	inhibition	DB02659	CDH1~P12830	Cholic Acid	STITCH	ACTS_ON	0.957
4	Calcitriol is an active metabolite of vitamin ...	inhibition	DB00136	CDH17~Q12864	Calcitriol	STITCH	ACTS_ON	0.957
5	A complex of related glycopeptide antibiotics ...	inhibition	DB00290	LIG3~P49916	Bleomycin	STITCH	ACTS_ON	0.958
6	Tolbutamide is an oral antihyperglycemic agent...	inhibition	DB01124	GCG~P01275	Tolbutamide	STITCH	ACTS_ON	0.957
7	Vildagliptin, previously identified as LAF237,...	inhibition	DB04876	GCG~P01275	Vildagliptin	STITCH	ACTS_ON	0.957
8	Glyburide is a second generation sulfonylurea ...	inhibition	DB01016	GCG~P01275	Glyburide	STITCH	ACTS_ON	0.957
9	None	inhibition	DB04297	HDAC1~Q13547	Trichostatin A	STITCH	ACTS_ON	0.938
10	Vorinostat (rINN) or suberoylanilide hydroxami...	inhibition	DB02546	HDAC1~Q13547	Vorinostat	STITCH	ACTS_ON	0.987
11	Gemcitabine is a nucleoside analog used as che...	inhibition	DB00441	CMPK1~P30085	Gemcitabine	STITCH	ACTS_ON	0.991
12	Resveratrol (3,5,4'-trihydroxystilbene) is a p...	inhibition	DB02709	PTGES~O14684	Resveratrol	STITCH	ACTS_ON	0.957
13	None	inhibition	DB03222	TXN~P10599	dATP	STITCH	ACTS_ON	0.900
14	Atorvastatin (Lipitor®), is a lipid-lowering d...	inhibition	DB01076	THBS1~P07996	Atorvastatin	STITCH	ACTS_ON	0.957

treatment_list = ['"{}"'.format(t) for t in treatment_regimens['treatment'].tolist()]
proposed_list = ['"{}"'.format(t) for t in drugs_proposed["drug_id"].unique()]
q = selected_queries.loc['side_effects_jaccard_similarity_intervention_proposed_drugs', 'query'].replace("INTERVENTION",",".join(treatment_list)).replace("DRUG",",".join(proposed_list))
similarity = connector.getCursorData(driver=driver, query=q, parameters={})

q

'MATCH (d1:Drug)-[:HAS_SIDE_EFFECT]->(phenotype1) WHERE d1.name in ["Oxaliplatin","Capecitabine","Folinic acid","Fluorouracil"] WITH d1, collect(id(phenotype1)) as treatmentData MATCH (d2:Drug)-[:HAS_SIDE_EFFECT]->(phenotype2) WHERE d2.id IN ["DB00678","DB00675","DB01229","DB02659","DB00136","DB00290","DB01124","DB04876","DB01016","DB04297","DB02546","DB00441","DB02709","DB03222","DB01076"] AND d1 <> d2 WITH d1, treatmentData, d2, collect(id(phenotype2)) as proposedData, size(apoc.coll.intersection(treatmentData, collect(DISTINCT id(phenotype2)))) AS intersection RETURN d1.name AS from, d2.name AS to, intersection, gds.alpha.similarity.jaccard(proposedData, treatmentData) AS similarity ORDER BY similarity DESC\n'

similarity = similarity[(similarity['from'].isin(treatment_regimens['treatment'].tolist())) & (similarity['to'].isin(side_effects['node1'].unique().tolist()))]
similarity = similarity.groupby('to').mean().sort_values(by='similarity')

similarity

	intersection	similarity
to
Cholic Acid	5.666667	0.041012
Tolbutamide	11.000000	0.083821
Vildagliptin	15.000000	0.098243
Calcitriol	25.333333	0.156174
Vorinostat	26.000000	0.162747
Bleomycin	28.666667	0.192705
Glyburide	36.000000	0.219190
Tamoxifen	44.000000	0.223948
Gemcitabine	47.000000	0.256839
Atorvastatin	61.000000	0.277987
Losartan	61.000000	0.310073
Paclitaxel	97.000000	0.378702

similarity = similarity[similarity['similarity']<0.2]

similarity

	intersection	similarity
to
Cholic Acid	5.666667	0.041012
Tolbutamide	11.000000	0.083821
Vildagliptin	15.000000	0.098243
Calcitriol	25.333333	0.156174
Vorinostat	26.000000	0.162747
Bleomycin	28.666667	0.192705

5) Check in Literature Known Links Between Disease and Drugs

selected_queries = query_utils.find_queries_involving_nodes(queries=queries, nodes=["Disease", "Drug"], print_pretty=True)

selected_queries.head()

	Name	Description	involved_nodes	involved_rels	query	example
id
association_drug_disease_publication	drug co-mentioned with disease in publication	Return the list of publications co-mentioning ...	Drug,Disease,Publication	MENTIONED_IN_PUBLICATION	MATCH (drug:Drug)-[r:MENTIONED_IN_PUBLICATION]...	drugs = ['DB00439', 'DB06196']\ndisease = ['D...
association_combination_drug_disease_publication	combinations of drugs co-mentioned with disease	Return the list of publications co-mentioning ...	Drug,Disease,Publication	MENTIONED_IN_PUBLICATION	MATCH (drug:Drug)-[r:MENTIONED_IN_PUBLICATION]...	drugs = ['DB00439', 'DB06196']\ndisease = ['D...

reduced_drug_list = drugs_proposed.loc[drugs_proposed["node2"].isin(similarity.index.tolist() + no_registerd_side_effects), "drug_id"]
reduced_drug_list = ['"{}"'.format(d) for d in reduced_drug_list]
drug_disease_query = selected_queries.loc['association_drug_disease_publication', 'query'].format(",".join(reduced_drug_list),",".join(diseases))

drugs_publications = connector.getCursorData(driver=driver, query=drug_disease_query, parameters={})

drugs_publications.head()

	Year	linkout	node1	node2	type
0	1988	https://www.ncbi.nlm.nih.gov/pubmed/2454581	dATP	PMID:2454581	MENTIONED_IN_PUBLICATION
1	1992	https://www.ncbi.nlm.nih.gov/pubmed/1562461	Calcitriol	PMID:1562461	MENTIONED_IN_PUBLICATION
2	1981	https://www.ncbi.nlm.nih.gov/pubmed/7236488	Bleomycin	PMID:7236488	MENTIONED_IN_PUBLICATION
3	2007	https://www.ncbi.nlm.nih.gov/pubmed/17583686	Resveratrol	PMID:17583686	MENTIONED_IN_PUBLICATION
4	2009	https://www.ncbi.nlm.nih.gov/pubmed/19760127	Trichostatin A	PMID:19760127	MENTIONED_IN_PUBLICATION

net = viz.get_network(data=drugs_publications, identifier="drugs_publications", args={"source":"node1", "target":"node2", "node_size":"degree","title":"Proposed drugs", "color_weight":False})

utils.json_network_to_gml(net['net_json'], path='drugs_publications.gml')

drugs_publications.groupby('node1')["node2"].count().plot.bar();

drug_combinations_query = selected_queries.loc['association_combination_drug_disease_publication', 'query'].format(",".join(reduced_drug_list),",".join(reduced_drug_list),",".join(reduced_drug_list))

selected_queries.loc['association_combination_drug_disease_publication', 'query']

'MATCH (drug:Drug)-[r:MENTIONED_IN_PUBLICATION]-(publication:Publication) WHERE (drug.id IN [{}]) WITH publication, count(r) AS r_count WHERE r_count>1 MATCH (drug:Drug)-[r:MENTIONED_IN_PUBLICATION]-(publication)-[:MENTIONED_IN_PUBLICATION]-(drug2:Drug) WHERE (drug.id IN [{}]) AND drug2.id IN [{}] AND drug.id<>drug2.id RETURN (drug.name +", "+drug2.name) AS node1, publication.id AS node2, publication.linkout AS linkout, publication.year as Year, type(r) AS type\n'

drugs_combinations = connector.getCursorData(driver=driver, query=drug_combinations_query, parameters={})

drugs_combinations.head()

	Year	linkout	node1	node2	type
0	2010	https://www.ncbi.nlm.nih.gov/pubmed/20371703	Cholic Acid, Calcitriol	20371703	MENTIONED_IN_PUBLICATION
1	2010	https://www.ncbi.nlm.nih.gov/pubmed/20371703	Calcitriol, Cholic Acid	20371703	MENTIONED_IN_PUBLICATION
2	2019	https://www.ncbi.nlm.nih.gov/pubmed/31661763	Cholic Acid, Resveratrol	31661763	MENTIONED_IN_PUBLICATION
3	2019	https://www.ncbi.nlm.nih.gov/pubmed/31661763	Resveratrol, Cholic Acid	31661763	MENTIONED_IN_PUBLICATION
4	2016	https://www.ncbi.nlm.nih.gov/pubmed/26903812	Cholic Acid, Calcitriol	26903812	MENTIONED_IN_PUBLICATION

drugs_combinations.groupby('node1')["node2"].count().plot.bar(figsize=(15,3));

6) Proposed Drug Candidates:

• Vorinostat (https://www.drugbank.ca/drugs/DB02546)

---

Trichostatin A (https://www.drugbank.ca/drugs/DB04297)

import pandas as pd
drug_result = drugs_proposed[drugs_proposed['node2'].isin(['Trichostatin A','Vorinostat'])]
drug_result.columns = ["Drug_desc","Action","Drug_id","Protein","Drug_name","Drug_protein_source","Drug_interaction_type","Inhibition_score"]
drug_result['Side_effects'] = ";".join(side_effects[side_effects['node1']==drug_result["Drug_name"].tolist().pop()]['node2'].tolist())
drug_result['Publications'] = ";".join(drugs_publications[drugs_publications['node1']==drug_result['Drug_name'].tolist().pop()]['node2'].tolist())
protein_result = proteins_associated_lung_cancer[proteins_associated_lung_cancer['node1']==drug_result["Protein"].tolist().pop()]
protein_result.columns = ["Protein", "Disease", "Protein_disease_source", "Protein_disease_association_type", "Disease_score"]
result = pd.merge(drug_result, protein_result, on='Protein')
result = result[["Protein", "Disease",
                 "Protein_disease_source", "Protein_disease_association_type",
                 "Disease_score", "Drug_name", "Drug_id", "Drug_desc", "Action", "Drug_protein_source","Drug_interaction_type","Inhibition_score",
                 "Side_effects", "Publications"
                ]]

result

	Protein	Disease	Protein_disease_source	Protein_disease_association_type	Disease_score	Drug_name	Drug_id	Drug_desc	Action	Drug_protein_source	Drug_interaction_type	Inhibition_score	Side_effects	Publications
0	HDAC1~Q13547	lung cancer	DISEASES	ASSOCIATED_WITH	2.51	Trichostatin A	DB04297	None	inhibition	STITCH	ACTS_ON	0.938	Alopecia of scalp;Erythroderma;Syncope;Diarrhe...	PMID:32904337;PMID:31772153;PMID:27743148;PMID...
1	HDAC1~Q13547	lung cancer	DISEASES	ASSOCIATED_WITH	2.51	Vorinostat	DB02546	Vorinostat (rINN) or suberoylanilide hydroxami...	inhibition	STITCH	ACTS_ON	0.987	Alopecia of scalp;Erythroderma;Syncope;Diarrhe...	PMID:32904337;PMID:31772153;PMID:27743148;PMID...

regulation_result = regulation_table[regulation_table["identifier"]==drug_result["Protein"].tolist().pop()]
regulation_result

	-log10 pvalue	FC	Method	T-statistics	correction	dfk	dfn	effsize	efftype	group1	...	identifier	log2FC	mean(group1)	mean(group2)	padj	pvalue	rejected	s0	std(group1)	std(group2)
1885	0.782856	10.367528	SAMR Two class paired	-1.008813	permutation FDR (4.0 perm)	1	2	1.227	hedges	CANCER	...	HDAC1~Q13547	3.374	30.666	27.292	0.036735	0.164871	True	2	0.301	2.201

1 rows × 21 columns

References

• http://disease-ontology.org/

• http://stitch.embl.de/

• https://www.ncbi.nlm.nih.gov/pubmed/

• http://sideeffects.embl.de/

• https://www.drugbank.ca

• https://cytoscape.org/

• https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6068348/

• https://neo4j.com/