SemMedDB

Conecting to and using the SemMedDB MySQL container on rtxdev

Starting the necisary containers from a stopped state

First, make sure that rtxdev.saramsey.org is running then ssh into it and run the following commands:

sudo docker start semmed
sudo docker start umls
sudo docker start semrep
sudo docker exec -d semrep /semrep/public_mm/bin/skrmedpostctl start
sudo docker exec -d semrep /semrep/public_mm/bin/wsdserverctl start

Then wait a few seconds for everything to initialize and you should be good to go.

Using SemMedInterface.py

NOTE: Currently all functions return the response as a pandas dataframe with column headers

Establishing a Connection

After importing the SemMedInterface class (from SemMedInterface import SemMedInterface) all you need to do to esablish a connection with the SemMedDB, UMLS, and semrep containers on rtxdev.saramsey.org is run the following:

smdbi = SemMedInterface()

Retreaving info on relationships between a subject and object

If you want to get information about the edges between a subject with the cui C1234567 and a object with the cui C7654321 run the following:

smdbt.get_edges_between_snodes('C0000001','Name1','C0000002','Name2')

This will return a pandas dataframe with the columns: PMID, SUBJECT_NAME, PREDICATE, OBJECT_NAME

If you want diffent columns in the response add them as a list of strings using the result_col option:

smdb.get_edges_between_nodes('C0000001','Name1','C0000002','Name2',result_col = ['the', 'columns', 'you', 'want'])

The available columns are:

• PMID
• PREDICATE
• SUBJECT_CUI
• SUBJECT_NAME
• SUBJECT_SEMTYPE
• OBJECT_CUI
• OBJECT_NAME
• OBJECT_SEMTYPE

If you want to specify the predicate connecting the two nodes use the predicate option:

smdb.get_edges_between_nodes('C0000001','Name1','C0000002','Name2',predicate = 'ISA')

If you don't want to specify if either node is a subject or object you can do so by using the bidirectional option:

smdb.get_edges_between_nodes('C0000001','Name1','C0000002','Name2',bidirectional = True)

Getting connections between nodes with multiple hops

If you want to get all of the connections between node with a secific number of pivots use the following:

smdb.get_edges_between_subject_object_with_pivot('C0000001','Name1','C0000002','Name2',pivots = 5)

If you want the shortest path between a subject and object (under a maximum length) instead use the following:

smdb.get_short_paths_between_subject_object('C0000001','Name1','C0000002','Name2',max_length = 3)

Converting cuis back into node identifiers from KG

If you want the curie id for a cui simply run the following command:

smdb.get_node_from_cui('C0000001')

If instead you wish to get the name of the node simply add True as a second argument:

smdb.get_node_from_cui('C0000001', True)