Algorithms for Molecular Biology

Table 8 Empirical study results

From: Non-parametric and semi-parametric support estimation using SEquential RESampling random walks on biomolecular sequences

Dataset	PR-AUC (%)		ROC-AUC (%)
Dataset	GUIDANCE1	SERES + GUIDANCE1	GUIDANCE1	SERES + GUIDANCE1
IGIA	62.67	69.28	89.50	91.62
IGIB	73.60	87.47	94.49	97.39
IGIC2	72.67	75.36	82.25	83.87
IGID	63.74	76.30	95.10	96.73
IGIE	93.56	95.42	90.08	93.30
IGIIA	73.03	83.06	86.49	96.45
PA23	98.54	99.41	82.59	93.63
PE23	98.44	99.27	94.75	97.41
PM23	97.53	98.48	94.20	96.44
SA16	99.72	99.86	91.07	95.57
SA23	98.35	99.24	81.76	92.18

Dataset	PR-AUC (%)		ROC-AUC (%)
Dataset	GUIDANCE2	SERES + GUIDANCE2	GUIDANCE2	SERES + GUIDANCE2
IGIA	67.4	68.49	91.38	91.94
IGIB	80.66	86.72	96.47	97.38
IGIC2	74.44	73.27	84.63	82.51
IGID	75.15	78.38	96.44	97.09
IGIE	94.6	95.44	91.84	93.49
IGIIA	78.16	85.09	94.50	96.82
PA23	99.24	99.53	91.48	94.88
PE23	99.07	99.34	96.72	97.63
PM23	98.68	98.85	96.93	97.28
SA16	99.88	99.91	96.22	97.22
SA23	99.04	99.33	89.93	93.18

The empirical study made use of benchmark RNA datasets and curated reference alignments from the CRW database [3]. Results are shown for intronic (“IG” prefix) and non-intronic datasets (“P” prefix and “S” prefix, following “primary” and “seed” nomenclature from the CRW database). For each dataset, we report each method’s PR-AUC and ROC-AUC. For each dataset and pairwise method comparison, the best AUC is shown in italics. Methods, performance measures, table layout, and table description are otherwise identical to Table 5

Back to article page

ISSN: 1748-7188

Contact us

General enquiries: journalsubmissions@springernature.com