====== News ====== ==== Mar 08, 2012 ==== **WHESS.db** [[this>../../pph2/dbsearch.shtml|Quick search]] option added to the PolyPhen-2 web interface. ==== Feb 23, 2012 ==== **PolyPhen-2** build r394 was released which fixes a number of bugs reported since the initial v2.2.2 release. == ChangeLog: == * Added missing MultiZ alignments for hg18 UCSC genome assembly version * Corrected substitution profile scores for hg19 MultiZ alignments, which were calculated with the wrong PSIC parameters * Updated classifier models to reflect the changes in distribution of PSIC profile scores caused by the MultiZ updates To upgrade from the previous version, download the following two files and extract their contents over your existing installation: polyphen-2.2.2r394.tar.gz polyphen-2.2.2-alignments-multiz-2009_10.tar.bz2 ==== Feb 15, 2012 ==== **PolyPhen-2 v2.2.2** (r393) has been [[downloads#software|released]]. Web server was also brought up to date with the new software release, which includes UniProtKB/UniRef100 Release 2011_12 (14-Dec-2011), PDB/DSSP Snapshot 03-Jan-2012 (78,304 entries) and UCSC GRCh37/hg19 (08-Oct-2009) databases. The software release is accompanied by a new build of the whole human exome sequence space ([[downloads#datasets|WHESS]]) dataset of precomputed PolyPhen-2 annotations. == ChangeLog: == * Integrated bundled annotation and sequence databases * Integrated precomputed protein and genomic multiple alignments * Simplified installation (MySQL server no longer required) * Improved documentation, see INSTALL and README files included * Up to 150-fold increase in speed for large-scale surveys * Improved annotation coverage, especially for non-globular proteins * Makefile updated to provide automatic downloads of WEKA v3-6-6 and MAFFT v6.864 * Numerous bugs fixed No significant changes to the classifiers performance or prediction accuracy were introduced in this version ==== Aug 9, 2011 ==== Interim update of **MapSNPs** tool [[downloads|released]], with fixes for a number of earlier reported bugs: * Fixed reverse-order flanking nucleotides for transcripts on minus strand * Fixed incorrect CpG context annotations for transcripts on minus strand * Fixed BLAST search occasionally missing a valid UniProtKB match (-U option) Please see [[appendix_b|MapSNPs annotation summary]] description for the CpG context annotations changes. ==== May 8, 2011 ==== **Version 2.1.0** (r367) of the standalone **PolyPhen-2** software has been [[downloads|released]]. Web server is now updated to utilize new software release, as well as the protein sequences from UniProtKB/UniRef100 Release 2011_04 (05-Apr-2011) and structures from PDB/DSSP Snapshot 06-Apr-2011 (72,244 entries). New version improves significantly on the classifiers performance: ^ HumDiv classifier ^ v2.0.23r349 ^^^ v2.1.0r367 ^^^ | False Positive Rate | 5% | 10% | 20% | 5% | 10% | 20% | | True Positive Rate | 66% | 80% | 93% | 78% | 89% | 96% | | Probability Threshold | 0.987 | 0.841 | 0.067 | 0.953 | 0.432 | 0.024 | | Correctly Classified Instances | 86.7% ||| 89.8% ||| | Incorrectly Classified Instances | 13.3% ||| 10.1% ||| | Kappa statistic | 0.711 ||| 0.793 ||| ^ HumVar classifier ^ v2.0.23r349 ^^^ v2.1.0r367 ^^^ | False Positive Rate | 5% | 10% | 20% | 5% | 10% | 20% | | True Positive Rate | 40% | 58% | 76% | 53% | 68% | 83% | | Probability Threshold | 0.991 | 0.964 | 0.746 | 0.981 | 0.902 | 0.493 | | Correctly Classified Instances | 79.2% ||| 81.2% ||| | Incorrectly Classified Instances | 20.8% ||| 18.8% ||| | Kappa statistic | 0.568 ||| 0.625 ||| Note:\\ * **HumDiv** model uses 5% / 10% FPR thresholds for "probably damaging" / "possibly damaging" predictions * **HumVar** model uses 10% / 20% FPR thresholds for "probably damaging" / "possibly damaging" predictions You can also download fivefold cross-validation {{polyphen2rocs.pdf|ROC curves}} and {{polyphen2thresholds.pdf|FPR/TPR threshold plots}} for version 2.1.0, obtained on updated HumDiv and HumVar datasets, available for download [[downloads#datasets|here]]. == ChangeLog: == * Training sets and classifier models updated; new models show increased sensitivity and specificity * MSA pipeline configuration settings tuned to improve alignments quality and speed * Better handling of structural features for proteins with PDB 3D structures available * Support for Grid Engine and Platform LSF distributed computing environments added * MapSNPs installation procedure and documentation improved * Makefile updated to provide automatic downloads of WEKA v3-6-4 and MAFFT v6.853 * A number of (mostly minor) bugs fixed ==== Dec 9, 2010 ==== **Version 2.0.23** (r349) of the standalone **PolyPhen-2** software has been [[downloads|released]]. == ChangeLog: == * Integrated MapSNPs genomic SNP annotation tool now included, see ''INSTALL.mapsnps'' file for installation instructions * Experimental species-specific support added for mouse, rat, dog, and zebrafish proteomes * Classifier models updated. :!: New models are incompatible with PSIC profiles generated by previous versions of the software. Make sure you removed all old PSIC profile files (''*.prf'') from your ''scratch'' and ''precomputed'' directories before using new version * Several LEON memory allocation errors fixed, stability of the MSA pipeline improved * MapSNPs bug fixed which caused standard gene symbol missing from the annotations for non-coding SNPs * Makefile download links updated to WEKA v3-6-3 and MAFFT v6.843 * Version strings for all database sources are now written to XML report files