Limits...
ClaRNA: a classifier of contacts in RNA 3D structures based on a comparative analysis of various classification schemes.

Waleń T, Chojnowski G, Gierski P, Bujnicki JM - Nucleic Acids Res. (2014)

Bottom Line: Each doublet of spatially close ribonucleotide residues in a query structure is compared to clusters of reference doublets obtained by analysis of a large number of experimentally determined RNA structures, and assigned a score that describes its similarity to one or more known types of contacts, including pairing, stacking, base-phosphate and base-ribose interactions.The classifier can be easily extended to include new types of spatial relationships between pairs or larger assemblies of nucleotide residues.ClaRNA is freely available via a web server that includes an extensive set of tools for processing and visualizing structural information about RNA molecules.

View Article: PubMed Central - PubMed

Affiliation: Laboratory of Bioinformatics and Protein Engineering, International Institute of Molecular and Cell Biology in Warsaw, Warsaw 02-109, Poland Faculty of Mathematics, Informatics and Mechanics, University of Warsaw, Warsaw 02-097, Poland iamb@genesilico.pl.

Show MeSH
Comparison of the ClaRNA results for the testing set. Each value represents the percentage of doublets from the testing set detected by ClaRNA (CL), MC-Annotate (MC), RNAView (RV), FR3D (FR) and ModeRNA (MO). The number of canonical base pairs that were detected exclusively by ClaRNA was only 2516, as compared to the size of the test set of canonical base pairs (486 118). MC-Annotate, RNAView and FR3D returned 3060, 11 352 and 1684 assignments of canonical base pairs, respectively, that lacked support from other classifiers. The number of non-canonical base pairs (including wobble WW_cis UG/GU pairs) that were detected only by ClaRNA is 31 016 as compared to the size of the corresponding test set of non-canonical base pairs (246 290). MC-Annotate, RNAView and FR3D returned 31 592, 45 904 and 57 832 assignments of non-canonical base pairs, respectively, that lacked support from other classifiers. Furthermore, the number of stacking interactions detected by ClaRNA exclusively was 12 533 (the test set contains 721 851 doublets classified as ‘consensual’ by other methods). In comparison, MC-Annotate, ModeRNA and FR3D returned 0, 34 and 129 230 stacking assignments, respectively, that lacked support from other classifiers.
© Copyright Policy - creative-commons
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4231730&req=5

Figure 3: Comparison of the ClaRNA results for the testing set. Each value represents the percentage of doublets from the testing set detected by ClaRNA (CL), MC-Annotate (MC), RNAView (RV), FR3D (FR) and ModeRNA (MO). The number of canonical base pairs that were detected exclusively by ClaRNA was only 2516, as compared to the size of the test set of canonical base pairs (486 118). MC-Annotate, RNAView and FR3D returned 3060, 11 352 and 1684 assignments of canonical base pairs, respectively, that lacked support from other classifiers. The number of non-canonical base pairs (including wobble WW_cis UG/GU pairs) that were detected only by ClaRNA is 31 016 as compared to the size of the corresponding test set of non-canonical base pairs (246 290). MC-Annotate, RNAView and FR3D returned 31 592, 45 904 and 57 832 assignments of non-canonical base pairs, respectively, that lacked support from other classifiers. Furthermore, the number of stacking interactions detected by ClaRNA exclusively was 12 533 (the test set contains 721 851 doublets classified as ‘consensual’ by other methods). In comparison, MC-Annotate, ModeRNA and FR3D returned 0, 34 and 129 230 stacking assignments, respectively, that lacked support from other classifiers.

Mentions: We also compared the similarity of the results obtained by ClaRNA and other classifiers (Figure 3). Each value corresponding to the intersection of sets CL, MC, RV and FR represents the percentage of doublets from the testing set recognized by ClaRNA, MC-Annotate, RNAView and FR3D, respectively, with the same interaction type. Here, 100% is arbitrarily defined as the number of doublets classified into the same interaction type by at least two methods. This analysis illustrates the tendency of individual methods to agree with each other as well as to propose solutions that are at odds with classifications made by other methods. For classical base pairs, all methods showed excellent agreement with each other, with over 94% of the pairs classified in the same way by all methods tested. RNAView reports the highest number of classical base pairs that are not identified by other methods (2.57%). In the case of non-classical pairs, the agreement between all methods is much lower, close to 67%, and each method reports a sizeable fraction of pairs that are not classified as pairs by any other method. FR3D is particularly generous in reporting non-canonical pairs, as the number of such pairs identified by this method alone is equal to as much as 30.70% of cases where at least two methods agree with each other. For stacking interactions, MC-Annotate and ModeRNA are quite conservative, and FR3D is the only method to report over 16% of doublets as stacked, compared to the number of doublets classified as stacked by at least two methods considered. For base–phosphate interactions, ClaRNA agrees with FR3D, and for base–ribose interactions, ClaRNA reports those identified by FR3D, as well as identifies 18.37% additional ones that are not identified by FR3D.


ClaRNA: a classifier of contacts in RNA 3D structures based on a comparative analysis of various classification schemes.

Waleń T, Chojnowski G, Gierski P, Bujnicki JM - Nucleic Acids Res. (2014)

Comparison of the ClaRNA results for the testing set. Each value represents the percentage of doublets from the testing set detected by ClaRNA (CL), MC-Annotate (MC), RNAView (RV), FR3D (FR) and ModeRNA (MO). The number of canonical base pairs that were detected exclusively by ClaRNA was only 2516, as compared to the size of the test set of canonical base pairs (486 118). MC-Annotate, RNAView and FR3D returned 3060, 11 352 and 1684 assignments of canonical base pairs, respectively, that lacked support from other classifiers. The number of non-canonical base pairs (including wobble WW_cis UG/GU pairs) that were detected only by ClaRNA is 31 016 as compared to the size of the corresponding test set of non-canonical base pairs (246 290). MC-Annotate, RNAView and FR3D returned 31 592, 45 904 and 57 832 assignments of non-canonical base pairs, respectively, that lacked support from other classifiers. Furthermore, the number of stacking interactions detected by ClaRNA exclusively was 12 533 (the test set contains 721 851 doublets classified as ‘consensual’ by other methods). In comparison, MC-Annotate, ModeRNA and FR3D returned 0, 34 and 129 230 stacking assignments, respectively, that lacked support from other classifiers.
© Copyright Policy - creative-commons
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4231730&req=5

Figure 3: Comparison of the ClaRNA results for the testing set. Each value represents the percentage of doublets from the testing set detected by ClaRNA (CL), MC-Annotate (MC), RNAView (RV), FR3D (FR) and ModeRNA (MO). The number of canonical base pairs that were detected exclusively by ClaRNA was only 2516, as compared to the size of the test set of canonical base pairs (486 118). MC-Annotate, RNAView and FR3D returned 3060, 11 352 and 1684 assignments of canonical base pairs, respectively, that lacked support from other classifiers. The number of non-canonical base pairs (including wobble WW_cis UG/GU pairs) that were detected only by ClaRNA is 31 016 as compared to the size of the corresponding test set of non-canonical base pairs (246 290). MC-Annotate, RNAView and FR3D returned 31 592, 45 904 and 57 832 assignments of non-canonical base pairs, respectively, that lacked support from other classifiers. Furthermore, the number of stacking interactions detected by ClaRNA exclusively was 12 533 (the test set contains 721 851 doublets classified as ‘consensual’ by other methods). In comparison, MC-Annotate, ModeRNA and FR3D returned 0, 34 and 129 230 stacking assignments, respectively, that lacked support from other classifiers.
Mentions: We also compared the similarity of the results obtained by ClaRNA and other classifiers (Figure 3). Each value corresponding to the intersection of sets CL, MC, RV and FR represents the percentage of doublets from the testing set recognized by ClaRNA, MC-Annotate, RNAView and FR3D, respectively, with the same interaction type. Here, 100% is arbitrarily defined as the number of doublets classified into the same interaction type by at least two methods. This analysis illustrates the tendency of individual methods to agree with each other as well as to propose solutions that are at odds with classifications made by other methods. For classical base pairs, all methods showed excellent agreement with each other, with over 94% of the pairs classified in the same way by all methods tested. RNAView reports the highest number of classical base pairs that are not identified by other methods (2.57%). In the case of non-classical pairs, the agreement between all methods is much lower, close to 67%, and each method reports a sizeable fraction of pairs that are not classified as pairs by any other method. FR3D is particularly generous in reporting non-canonical pairs, as the number of such pairs identified by this method alone is equal to as much as 30.70% of cases where at least two methods agree with each other. For stacking interactions, MC-Annotate and ModeRNA are quite conservative, and FR3D is the only method to report over 16% of doublets as stacked, compared to the number of doublets classified as stacked by at least two methods considered. For base–phosphate interactions, ClaRNA agrees with FR3D, and for base–ribose interactions, ClaRNA reports those identified by FR3D, as well as identifies 18.37% additional ones that are not identified by FR3D.

Bottom Line: Each doublet of spatially close ribonucleotide residues in a query structure is compared to clusters of reference doublets obtained by analysis of a large number of experimentally determined RNA structures, and assigned a score that describes its similarity to one or more known types of contacts, including pairing, stacking, base-phosphate and base-ribose interactions.The classifier can be easily extended to include new types of spatial relationships between pairs or larger assemblies of nucleotide residues.ClaRNA is freely available via a web server that includes an extensive set of tools for processing and visualizing structural information about RNA molecules.

View Article: PubMed Central - PubMed

Affiliation: Laboratory of Bioinformatics and Protein Engineering, International Institute of Molecular and Cell Biology in Warsaw, Warsaw 02-109, Poland Faculty of Mathematics, Informatics and Mechanics, University of Warsaw, Warsaw 02-097, Poland iamb@genesilico.pl.

Show MeSH