Computational comparison of common event-based differential splicing tools: practical considerations for laboratory researchers.

BACKGROUND: Computational tools analyzing RNA-sequencing data have boosted alternative splicing research by identifying and assessing differentially spliced genes. However, common alternative splicing analysis tools differ substantially in their statistical analyses and general performance. This report compares the computational performance (CPU utilization and RAM usage) of three event-level splicing tools; rMATS, MISO, and SUPPA2. Additionally, concordance between tool outputs was investigated. RESULTS: Log-linear relations were found between job times and dataset size in all splicing tools and all virtual machine (VM) configurations. MISO had the highest job times for all analyses, irrespective of VM size, while MISO analyses also exceeded maximum CPU utilization on all VM sizes. rMATS and SUPPA2 load averages were relatively low in both size and replicate comparisons, not nearing maximum CPU utilization in the VM simulating the lowest computational power (D2 VM). RAM usage in rMATS and SUPPA2 did not exceed 20% of maximum RAM in both size and replicate comparisons while MISO reached maximum RAM usage in D2 VM analyses for input size. Correlation coefficients of differential splicing analyses showed high correlation (ß > 80%) between different tool outputs with the exception of comparisons of retained intron (RI) events between rMATS/MISO and rMATS/SUPPA2 (ß < 60%). CONCLUSIONS: Prior to RNA-seq analyses, users should consider job time, amount of replicates and splice event type of interest to determine the optimal alternative splicing tool. In general, rMATS is superior to both MISO and SUPPA2 in computational performance. Analysis outputs show high concordance between tools, with the exception of RI events.

On the design and analysis of competent selecto-recombinative GAs.

In this paper, we study two recent theoretical models--a population-sizing model and a convergence model--and examine their assumptions to gain insights into the conditions under which selecto-recombinative GAs work well. We use these insights to formulate several design rules to develop competent GAs for practical problems. To test the usefulness of the design rules, we consider as a case study the map-labeling problem, an NP-hard problem from cartography. We compare the predictions of the theoretical models with the actual performance of the GA for the map-labeling problem. Experiments show that the predictions match the observed scale-up behavior of the GA, thereby strengthening our claim that the design rules can guide the design of competent selecto-recombinative GAs for realistic problems.