Connecting to ontox4j

Connect and query the ontox4j graph database. TODO don’t use distill

Install & Connect

We need neo4j for python, R doesn’t have an active driver. So we use reticulate:

1. Install reticulate install.packages("reticulate")
2. Create a conda environment reticulate::conda_install()
3. Install neo4j pip install neo4j

Connect to ontox4j with the python driver. First install reticulate

neo4j  <- reticulate::import("neo4j")

Then use the very simple python driver wrapper below. We can add this to an eventual package ():

from neo4j import GraphDatabase
class ontox4j_driver: 
  
    def __init__(self, uri, user, password):
        self.driver = GraphDatabase.driver(uri, auth=(user, password))

    def close(self):
        self.driver.close()

    def query(self,query):
        with self.driver.session() as session:
            tx   = session.begin_transaction()
            return [row for row in tx.run(query)]

Create some proteins

Ontox4j already has a few chemical relationships. Create a cypher query to view them.

con = reticulate::py$ontox4j_driver(Sys.getenv("ontox4j_uri"),Sys.getenv("ontox4j_usr"),Sys.getenv("ontox4j_pw"))


create_node  = \(tag,...){ # Create nodes
  node    = list(...)[-1]
  keyvals = glue::glue("{names(node)}:'{node}'") |> paste(collapse=",")
  query   = glue::glue("CREATE (x:{tag} {{ {keyvals} }}) RETURN x")
  con$query(query)[[1]]
}

nodes     = list(
  RALDH2    = create_node("Protein" ,name="RALDH2",type="enzyme",uniprot="094788"),
  RDH10     = create_node("Protein", name='RDH10',type='enzyme',uniprot='Q8IZV5'),
  VIT_A_ext = create_node("Compound",name='Vitamin A_e',type="vitamin",pubchem_cid="445354",location="extracellular"),
  VIT_A_int = create_node("Compound",name='Vitamin A_i',type='vitamin',pubchem_cid='445354',location='intracellular'),
  ATRA      = create_node("Compound",name= 'ATRA',type= 'metabolite',pubchem_cid= '444795',location= 'intracellular'),
  Retinald  = create_node("Compound",name='Retinaldehyde',type='metabolite',pubchem_cid='638015'))


create_edge = \(rtag,tag1,name1,tag2,name2){ # create relationships TODO use create_edge(node1, "relationship", node2)
  query = glue::glue("MATCH (n1:{tag1}),(n2:{tag2}) WHERE n1.name='{name1}' AND n2.name='{name2}'
                     CREATE (n1)-[r:{rtag}]->(n2) 
                     RETURN n1, n2, r")
  con$query(query)
}

`vitai->rdh10` = create_edge("metabolizes","Compound","Vitamin A_i","Protein","RDH10")
`vitae->rdh10` = create_edge("translocates","Compound","Vitamin A_e","Protein","STRA6")

`vitai->reti`  = create_edge("is_converted_to","Compound","Vitamin A_i","Compound","Retinaldehyde")
`vitae->vitai` = create_edge("translocated_from","Compound","Vitamin A_e","Compound","Vitamin A_i")

`raldh->reti`  = create_edge("metabolizes","Protein","RALDH2", "Compound","Retinaldehyde")

query the results

nodes = con$query("MATCH (n) RETURN n") |> purrr::map(~ .$data()$n) |> 
  purrr::transpose() |> tibble::as_tibble() |> 
  tidyr::unnest(everything())

# TODO some work to do here
edges = con$query("MATCH (n) MATCH (n)-[r]->(m) RETURN n,r,m;")

clean up

con$query("MATCH (n) DETACH DELETE n") # This doesn't work, not sure why.

list()

con$close()