For mRNA-seq data integrated in RhesusBase, nine parameters were considered in the evaluation and scoring of NGS datasets, including:

1. normalized RNA integrity number as an indicator for the quality of RNA samples;
2. normalized lengths of RNA-seq reads;
3. normalized percentage of high quality sites (the 25th percentile Phred quality score ≥ 20) across all bases of the reads;
4. the quality of strand specificity, estimated on the basis of the percentage of reads with correct strand assignment. For RNA-seq datasets with strand non-specific design, this sub-score was set to zero;
5. the quality of short reads mapping efficiency, estimated on the basis of the percentage of uniquely-mapped reads to the reference genome;
6. the mutation rates across the reads, estimated by the percentage of sites with low mutation rate (≤0.01) across the reads;
7. degrees of potential contamination. To ensure the RNA-seq reads were generated from mature mRNAs, the RPKM of exonic, intronic and intergenic regions were calculated and compared, and the ratios of RPKMexonic/RPKMintronic and RPKMexonic/RPKMintergenic were normalized and summarized to estimate the degrees of contamination for one RNA-seq study;
8. the coverage of the transcriptome, estimated by the percentage of sites with high reads coverage (≥ 10) across the whole transcriptome;
9. the quality of fragmentation indicated by the distribution of reads on the transcript. Each transcript was equally divided into 25 intervals. For an ideal RNA-seq performed with perfect RNA fragmentation, the fraction for RNA-seq reads mapped into each interval is expected to 0.04. When combining all transcripts across the transcriptome, the absolute deviations to 0.04 for each interval were summed and normalized to indicate the quality of fragmentation in one RNA-seq study.