1. Load the spark-shell:

        $ spark-shell

2. Import the graphx related classes:

        scala> import org.apache.spark.graphx._

3. Load the edges from hdfs:

        scala> val edgesFile = sc.textFile(
          "hdfs://localhost:9000/user/hduser/data/cc/edges.csv")

4. Convert the edgesFile RDD into the RDD of edges:

        scala> val edges = edgesFile.map(_.split(",")).map(e => 
          Edge(e(0).toLong,e(1).toLong,e(2)))

5. Load the vertices from HDFS:

        scala> val verticesFile = sc.textFile(
          "hdfs://localhost:9000/user/hduser/data/cc/nodes.csv")

6. Map the vertices:

        scala> val vertices = verticesFile.map(_.split(",")).map( e => 
          (e(0).toLong,e(1)))

7. Create the graph object:

        scala> val graph = Graph(vertices,edges)

8. Calculate the connected components using graph.connectedComponents:

        scala> val cc = graph.connectedComponents

9. Find the vertices for the connected components (which is a subgraph):

        scala> val ccVertices = cc.vertices

10. Print the ccVertices:

       scala> ccVertices.collect.foreach(println)
       (4,4)
       (6,4)
       (2,1)
       (1,1)
       (3,1)
       (5,4)

As you can see in the output, vertices 1, 2, and 3 are pointing to 1, while 4, 5, and 6 are pointing to 4. Both of these are the lowest-indexed vertices in their respective clusters.