ak-graph

AK Graph Algorithms

Compile and setup

• git clone https://github.com/airr-knowledge/ak-graph.git
• CD into ak-graph and init submodules: git submodule update --init --recursive
• cp .env.defaults .env
• nano .env # set database and other deployment config
• Run make to show help.
• Most make targets are performed within a docker container.
• Run make docker to build docker image. The docker will compile all programs.

docker info

• Compiled programs will be in /ak-graph directory inside docker, but use /work for development.
• /ak-graph/bin and /work/bin are added to PATH

development, start shell inside docker, graph data at /ak_graph_data is stored in work

• Run docker run --network ak-db-network -v $PWD:/work -v ./graph_data:/ak_graph_data -it airrknowledge/ak-graph bash to start ak-graph container.
• Create a bash alias to help alias docker-ak-graph-dev="docker run --network ak-db-network -v $PWD:/work ./graph_data:/ak_graph_data -it airrknowledge/ak-graph bash"

production runs

• GRAPH_DATA environment variables is the host folder for the ak-graph data

• Run docker run --network ak-db-network -v ${GRAPH_DATA}:/ak_graph_data -it airrknowledge/ak-graph bash for shell in container.

• Run make create-all-table to generate unique CDR3 sequence tables in the database.

• Run make cdr3-edgelist-threads to extract CDR3 sequences from database and generate distance=1 graph.

• Run make build-graph to generate the graph from the edge list and save it in networkit format.

• Run make graph-stat to perform some general statitical analysis on graph.

• Run make epitope-analysis to perform epitope analysis using available IEDB epitopes in conjunction with the junction_aa graph.

Timing Analysis

Program	Threads	CHUNK	real	user	sys	query_and_stream()	run_compairr()	db_read()	compute_overlap()
stream_query	1	1000000	117m53.018s	110m32.805s	7m14.195s	0m38.7467s	117m14.116s	0m29.2801s	116m43.7309s
stream_query_threads	16	1000000	81m45.693s	128m23.770s	7m56.293s	0m39.0097s	81m5.9671s	0m31.7337s	80m34.2332s
stream_query_threads	48	1000000	91m28.135s	156m2.117s	17m14.116s	0m39.2061s	90m48.3653s	0m29.2814s	90m19.0837s
stream_query_threads	64	1000000	94m53.812s	182m44.747s	16m26.075s	0m38.8788s	94m14.318	0m29.0449s	93m14.318s
stream_query_threads	96	1000000	93m55.463s	241m57.294s	13m48.724s	0m39.0387s	93m15.6882s	0m28.963s	92m46.7249s
No Output
stream_query_no_output	1	1000000	51m18.046s	50m21.783s	0m51.085s	0m41.3267s	50m36.1499s	0m29.4679s	50m6.68193s
stream_query_threads_no_output	1	1000000	51m25.474s	50m37.785s	0m42.631s	0m38.9758s	50m46.0623s	0m29.0679s	50m16.9943s
strean_query_threads_no_output	2	1000000	32m5.100s	59m21.272s	3m5.270s	0m39.1839s	31m25.4246s	0m29.4821s	30m55.9424s
stream_query_threads_no_output	4	1000000	15m5.802s	53m24.392s	1m33.136s	0m42.0846s	14m23.194s	0m30.0385s	13m53.1553s
stream_query_threads_no_output	8	1000000	9m37.324s	63m19.513s	1m16.326s	0m39.4549s	8m57.3962s	0m29.5826s	8m27.8135s
stream_query_threads_no_output	16	1000000	6m28.831s	75m51.107s	1m9.708s	0m39.3275s	5m48.8363	0m29.7976s	5m19.0386s
stream_query_threads_no_output	32	1000000	4m42.837	97m36.692s	1m30.149s	0m39.2919s	4m3.04755s	0m29.27s	3m33.7774s
stream_query_threads_no_output	48	1000000	4m25.823s	124m0.592s	1m38.509s	0m38.9672s	3m46.2524s	0m29.8667s	3m16.3855s
stream_query_threads_no_output	64	1000000	4m18.874s	160m34.867	1m32.924s	0m39.1994s	3m39.1776s	0m29.6685s	3m9.50895s
stream_query_threads_no_output	96	1000000	4m3.507s	207m18.392s	1m55.971s	0m39.5649s	3m23.421s	0m29.4086s	2m54.0122s