Fig 1.
A) Cartoon representation of the Sex-lethal RRM1 protein in complex with polyU (PDB Id. 1b7f). The main secondary structure elements are labelled and coloured in pale green (β-strands), dark salmon (⍺-helices) and light blue (loops and terminal regions). B) Schematic representation of the RRM depicting the main positions of the canonical RNA binding interface with single circles and highlighting in red the most significant ones in terms of RNA interactions prevalence. The same colour code is used and the light green and dark green circles in the β-strands correspond to the exposed and buried residue sidechains, respectively.
Fig 2.
Data flow diagram for the RRM structural and sequence data to generate the master alignment, use it to align the RRM-RNA complexes and cluster them depending on their binding mode so the RNAs can be aligned.
The different Datasets that have been made available are named and labelled with its corresponding number of entries.
Fig 3.
Schematic representation of the sequence alignment procedure to improve the alignment in the loop and terminal regions.
Fig 4.
Percentage of contacts (Blue line) and gaps (grey dashed line) for all the positions of the 271 bound RRM alignment.
The β-strands and α-helices are depicted and labelled in green and red, respectively.
Fig 5.
Gap fraction for the 5 core RNA positions in the alignment used for the scoring (top) and nucleotide conservation (bottom). Sequence logo generated with WebLogo [38].
Fig 6.
Score distribution for the training set (scores from experimentally solved RRM-RNA complexes) and randomized set (randomly generated RNAs).
Fig 7.
Correlation between the bits values derived from the RNAcompete data and the scores obtained with RRMScorer.
For a clearer depiction of the single RRM (orange), multiple RRM (yellow) and multiple RBP (green) categories, each box corresponds to the distribution of the bits values for a range of 0.1 in the score axis.
Fig 8.
Experimental Kd values-scores correlation for the 36 RNA mutants tested against MSI1 RRM1, using a window size of three nucleotides as it is the length that this RRM specifically recognizes [12].
Fig 9.
A) Cartoon representation of the MSI1 RRM1 protein in complex with GUAGU (G1 not shown, PDB Id. 2rs2). The protein backbone is shown in light blue and heavy atoms are shown in red (O atoms), blue (N atoms), Orange (P atoms), light blue (C atoms of RRM) and sea blue (C atoms of RNA). The RNA nucleotides and the residues involved in the specific recognition of G4 (K21 and F65) are shown as sticks and spheres. B) and C) Score matrices for the alignment positions matching the interactions between K21 and F65 with G4, respectively (highlighted in bright green).
Fig 10.
Scores for the 5-nucleotide sliding windows of the WT RNA tested by N. Ruth Zearfoss et al.
The bars are coloured depending on the score confidence, that goes from 0 to 1 for no-confidence to maximum confidence predictions, respectively. To ease the interpretation of the results, we add 0.89 to each of the scores obtained as it is the value that better separates the training and randomized regions according to the receiver operating characteristic curve (ROC curve). Therefore, positive scores correspond to likely binders while negative scores correspond to RNA fragments less likely to bind the RRM.
Fig 11.
A) Cartoon representation of the SRSF1 RRM1 protein in complex with AACAAA (PDB Id. 6hpj). The protein backbone is shown in pale green and heavy atoms are shown in red (O atoms), blue (N atoms), Orange (P atoms), pale green (C atoms of RRM) and pale orange (C atoms of RNA). The RNA nucleotides and the residue involved in the specific recognition of C3 (E87) are shown as sticks and spheres. The interaction between E87 and C3 is shown as yellow dashed line. B) Score matrix for the respective alignment position of the interaction between E87 and C3. The change in score for the E87N mutation performed by Cléry et al. is highlighted in bright green.
Table 1.
Scores for the 4 tested RNAs by A. Cléry et al. coloured in green for the RNAs that bind the target on their NMR assays and in red for the RNAs that do not bind. The symbols reflect the score change after the E87N mutation.