Cloud-scale RNA-sequencing differential expression analysis with Myrna

Table 1 Myrna runtime, cost for 1.1 billion reads from the Pickrell et al. study [32]

EC2 nodes	1 master, 10 workers	1 master, 20 workers	1 master, 40 workers
Worker processor cores	80	160	320
Wall clock time	4h:20m	2h:32m	1h:38m
Cluster setup	4m	4m	3m
Align	2h:56m	1h:31m	54m
Overlap	52m	31m	16m
Normalize	6m	7m	6m
Statistics	9m	6m	6m
Summarize and Postprocess	13m	14m	13m
Approximate cost (location dependant)	$44.00/$49.50	$50.40/$56.70	$65.60/$73.80

Timing and cost for a Myrna experiment with 1.1 billion 35-bp unpaired reads from the Pickrell et al. study [32] as input. Costs are approximate and based on the pricing as of this writing, that is, $0.68 per High-CPU Extra Large instance per hour in the Northern Virginia zone and $0.78 elsewhere, plus a $0.12 per-node-per-hour surcharge for Elastic MapReduce. The table does not include the transfer of the read data from University of Chicago servers to S3, which takes a wall clock time of 1h:15 m and costs about $12. Other charges may apply and times may vary subject to, for example, congestion and Internet traffic conditions.

ISSN: 1474-760X