PrionW predicts proteins containing Q/N rich prion domains (PrDs) and identifies the 21-residues amyloid cores nucleating their amyloid formation. We found that the potency of the amyloid core correlates significantly with the prionic propensities of Q/N rich sequences (Sabate et al. 2015, PLoS Comput Biol. 11(1):e1004013).
The algorithm runs as follows:
1) It uses FoldIndex to identify disordered regions.
2) It looks for Q/N rich segments in the identified disordered regions.
3) It uses pWALTZ to identify and score the amyloid cores in Q/N rich segments.
1) The user can select the minimum Q/N richness of the PrD he wants to identify. The default is set at ≥ 25%, as all known yeast prions fulfill this requirement. Because PrionW is aimed to identify Q/N rich PrDs, values < 10% are not recommended. PrionW is not intended to delimit the exact boundaries in the identified PrLDs; however the best overlap between Uniprot annotated prion domains for actual yeast prions and PrionW predictions was obtained when Q+N richness was set to ≥ 32%.
2) The user can select the pWALTZ cutoff. The default is set at 73.55, since this cutoff provides the best accuracy when analyzing the yeast proteome (94.1%). A lower cutoff can be useful to identify sequences in genomes with a basal prion propensity in which mutations can promote conversion to a prionic phenotype.
3) The user can paste up to 10000 sequences in the prediction window and/or upload a file containing the sequences. In both cases, a FASTA format is mandatory.
1) For any sequence in dataset PrionW will indicate immediately:
- The number of sequences it is analysing.
- The parameters the user has selected for the prediction.
- A job identification number.
2) If your dataset does not contain any positive hit, the following message will appear: None of your sequences contains a predicted Prion-like Domain. In any other circumstance the algorithm is running.
3) For each particular positive sequence the algorithm provides the following output:
- Name of the sequence.
- The 21 residues amyloid core.
- The pWALTZ score of the amyloid core.
- The predicted Q/N rich prion domain with the amyloid core highlighted in red.
4) The user can download the prediction as a .csv file containing all the above information.
If you want to try PrionW you can pre-populate the prediction window by clicking the Examples link in the front-page. The dataset contains the full-length sequences of the well-characterized yeast prions NEW1, RNQ1, SWI1, SUP35 and URE2 and a set of prion positive and negative control synthetic sequences as designed by Toombs et al. (2012, PNAS 109: 6519-6524).