From Chasm Software Wiki
Release History
Current Release
03/05/2014 - CHASM 3.0
- Classifiers were retrained with SNVBox 3.0.
03/05/2014 - SNVBox 3.0
- Constructed with the latest RefSeq, CCDS, and Ensembl transcripts.
03/05/2014 - VEST 3.0
- Updated the VEST training set to use HGMD2013v4 release.
- The classifier was retrained with SNVBox 3.0.
Previous Releases
09/19/2013 - CHASM 1.0.8
- Updated list of precomputed classifiers. Classifiers were trained with 13885 driver mutations which is ~9000 mutations more than in previous versions. New mutations are from COSMIC v65 and include all driver genes identified in Vogelstein et al., Cancer Genome Landscapes, Science. 2013 Mar 29;339(6127):1546-58.
- Passenger gene selection: If fewer than the specified number of passenger genes (specified in chasm_classifiers.conf) are mutated in the input list, the genes in the input list used in passenger generation are augmented with a random selection of genes, which are known to be cancer-associated.
- 26 classifiers are included: 20 context tables estimated from TCGA sequencing studies, 4 from ICGC studies, 1 from Johns Hopkins (Velculescu/Kinzler/Vogelstein lab), and 1 from NHGRI/NCI (Samuels lab). All features are generated with SNVBox 2.0.0.
09/19/2013 - SNVBox 2.0.0
- Constructed with the latest RefSeq, CCDS, and Ensembl transcripts. MUTAGEN feature has been removed. Updated with dbSNP 137.
09/19/2013 - VEST 1.1.0
- Updated the VEST training set to use HGMD2013v2 release
04/17/2013 - CHASM 1.0.7
- Updated list of precomputed classifiers. The list now includes classifiers for 5 new tissue types (Kidney, Bladder, Head and Neck, Lung Adenocarcinoma and Lung Squamous Cell) and an updated classifier for AML.
3/21/2012 - SNVBox 1.0.2
- Updated CodonTable to improve coverage for converting genomic coordinates to protein position and amino acid substitution.
06/28/2012 - CHASM Bugfix 1.0.6
- Fixed a minor bug in FDR bin calculation that was causing some p-values to fall under a more conservative FDR bin. P-value reporting was altered to provide 4 significant digits. Classifiers available through the CRAVAT website are now packaged with CHASM, under BuiltClassifiers.
06/28/2012 - CHASM 1.0.6
- For reproducibility, CodonMap.py was altered to take the first missense mapping in any case where a nucleotide substitution maps to multiple missense mutations. This can happen if 1) the codon onto which a nucleotide substitution is mapped is split across an exon boundary. In such cases, alternative splicing can result in multiple possible mappings. 2) In rare cases, it is also possible that transcripts are predicted to overlap the genomic coordinate on both the positive and negative DNA strands.
3/21/2012 - SNVBox 1.0.1
- Added the CodonTable to allow fast mapping of genomic coordinates onto transcript, codon position and amino acid substitution.
06/06/2012 - CHASM BugFix 1.0.5
- Fixed bug in snvGetGenomic causing improper feature retrieval for mutations submitted in genomic coordinates.
3/21/2012 - CHASM 1.0.5
Warning: this release had a bug in snvGetGenomic
- Converted from compiled binaries to executable python source code.
- Added support for genomic coordinates by splitting snvGet into snvGetTranscript, snvGetGenomic and snvGetTranscriptList
- Updated background passenger frequency tables in ClassifierPack/contexts and added tables for several new tumor types.
- Changed Random Forest software from Waffles Learn to Parf. Parf implements the Random Forest algorithm originally described by Leo Brieman.
1/25/2012 - CHASM 1.0.4
- Fixed glibc compatibility runtime issues for the 64-bit binaries
- Added better error messages when the program fails to connect to the MySQL database
11/11/2011 - CHASM 1.0.3
- Updated list of driver mutations. We have removed TP53 mutations that were shown to have equivalent activity to wild-type TP53 in a WAF1 transactivation assay. Data courtesy of Dr. Thierry Soussi (Karolinska Institute, Sweden).
- The null set generation method that is calibrated by the scores of a known set of severe driver mutations is no longer the default. We have found this option to be useful in improving classifier sensitivity when the number of mutations being classified is small (less than 100), but it may reduce classifier specificity. We do not recommend using it when the number of mutations being classified is large. If you wish to use the option, go to the CHASM/chasm_classifiers.conf file and set usecalibration=True
- The default for features used by the classifier is now all 86 features available in SNVBox. You can use any desired subset of features by creating a file of feature IDs and using the -f option on the command line. Details are here.
9/13/2011 - CHASM 1.0.2
- Updated list of driver mutations
- Unix sockets can now be used for local connections to MySQL server for SnvGet
- Context tables do not have to be tab-delimited, "-" can be used in place of "0"
- Silent mutations are now filtered out from the list of inputs
- Fixed SnvGet bug for handling multiple input files when not in class mode
7/27/2011 - CHASM 1.0.1
- There were some mistakes with the binary names in the 32-bit download packages CHASMDL32.tar.gz and SNVBOXDL32.tar.gz, which can cause the program to fail because it tries to look for CamelCase program names in the CHASM INSTALLATION DIRECTORY that are not there. This problem has been fixed in the latest download packages.
- If you are using 64-bit versions then you are not affected by this.
4/10/2011 - CHASM 1.0.0
- BuildClassifier program now requires an additional input file, the user's mutation list.
- Bug fixes for problems with P-value and FDR calculations applicable when number of mutations to be classified is small (<50).
- Modified null set generation method to be calibrated by the scores of a known set of severe driver mutations
- Modified FDR calculation so that FDR is not calculated if a very small number of mutations is submitted to be classified. Default is 10, but can be configured to any number desired.)
- Modified "white list" of genes from which synthetic passenger mutations can be generated. Original white list consisted of 9,000 genes that were somatically mutated in five cancer sequencing studies at Sidney Kimmel Cancer Center. The new white list consists of the genes in which the mutations submitted by the user are found, or if this list is less than a user-configurable threshold (default=400) the union of the original white list and the user-submitted genes.
- SnvGet is now modified so that users can retrieve original, non-normalized feature values.
- SnvGet is now modified so that a user can choose not to fill-in missing feature values.
- All program names are now camel-cased.
4/10/2011 - SNVBox 1.0.0
- 86 features
- Supports feature lookup based on user supplied transcript identifier, amino acid substitution and codon position.
- Supports RefSeq, Ensembl and CCDS transcript identifiers.
- Transcripts are mapped to genomic coordinates in GRCh37/hg19.
2/21/2011 - CHASM beta 0.1.2
- Optimized SNVget for retrieving large number of mutations
(Retrieval time about 15 minutes for 10000 mutations when tested on our lab machine. Actual time will differ based on MySQL server specifications.)
1/27/2011 - CHASM beta 0.1.1
- Fixed bug with Benjamini-Hochberg FDR correction in runchasm
11/24/2010 - CHASM beta 0.1
- Initial release