a lexical manipulation toolkit for genetic variant descriptors (hgvs, etc)
Python toolkit for parsing HGVS strings describing genetic variants (mutations) into components and generating lexical synonyms of these descriptors.
You’ll need a UNIX-like environment to use this package. Both OS X and Linux have been confirmed to work.
Additionally, your system must contain the C development version of the client library for Postgres 9.3 or later (needed to connect to UTA services).
On OS X using brew:
$ brew install postgresql
On Linux using Ubuntu:
$ apt-get install libpq-dev
Python library dependencies for variant_lexicon are the following:
numpy biopython uta hgvs
The above “should” be successfully installed by using setup.py and/or by installing this package via pip.
Install ipython via pypi:
$ pip install ipython
More installation options and instructions are available on the iPython installation page.
Variant is an API shortcut to instantiating a SequenceVariant object from an HGVS string. It accepts both gene-name-bearing and non-gene-name-bearing HGVS strings.
See [Human Genome Variation Society](http://www.hgvs.org/) for explanation of HGVS strings.
Examples of valid HGVS strings:
NM_198056.2:c.4786T>A NC_000023.10:g.32841489delT NP_009225.1:p.Glu1000_Glu1001del NM_001165963.1(SCN1A):c.384_1662dup
Note: some valid HGVS strings currently cannot be parsed by the hgvs library (as of 6/17/2016). Examples:
When an illogical or invalid HGVS string is provided to Variant, this function returns None. (Tune into the “metavariant” logger at INFO level for information.)
Variant will also accept a SequenceVariant object as its argument, returning the same variant (identity).
VariantLVG takes an hgvs.SequenceVariant object or a plain HGVS string and uses the Universal Transcript Archive (UTA) to find as many related transcripts and sequence variants as possible that could be used as “synonyms” of the provided variant.
This class raises CriticalHgvsError upon instantiation if the supplied HGVS string fails to create a valid SequenceVariant object.
hgvs_text = 'NM_198056.2:c.4786T>A' lex = VariantLVG(hgvs_text) print(lex.variants) print(lex.transcripts) print(lex.gene_name)
VariantLVG provides for “enrichment” of lexical variant generation by allowing more transcripts and variations to be supplied at instantiation. Just use the appropriate keyword for the type of information, remembering that the “enrichment” keyword arguments are all lists.
VariantComponents allows instantiation in two different ways: using an hgvs.SequenceVariant object and using a set of components as keyword arguments.
Usage starting from a SequenceVariant object:
comp = VariantComponents(seqvar)
Usage starting from individual components:
comp = VariantComponents(seqtype='c', edittype='SUB', pos='322', ref='C', alt='T')
If no seqtype is supplied, VariantComponents tries to infer the sequence type heuristically (e.g. the presence of a “U” in the ref or the alt implies this is an RNA string).
VariantComponents may raise a RejectedSeqVar exception during instantiation (see “Exceptions” below).
A VariantComponents object provides access to the following attributes and properties:
seqtype: the sequence type of this seqvar (one of 'c', 'g', 'g', 'n') edittype: the type of mutation represented by this variant ('SUB', 'DEL', 'FS', etc) pos: position of the edit ref: reference sequence at given position (aka "wildtype") alt: alternate (or "mutant") at given position posedit: returns the HGVS "official" construction of this seqvar's position + edit information. posedit_slang: returns a list of algorithmically generated "slang" for given seqvar's posedit.
All exceptions can be found and imported from metavariant.exceptions.
CriticalHgvsError: raised when input HGVS string fails to instantiate a SequenceVariant object within the VariantLVG object.
RejectedSeqVar: raised inside VariantComponents when input sequence components fail certain tests of completeness. For example, a protein seqvar will throw this Exception if the protein effect string is only a “?” (i.e. unknown protein effect). A “SUB” (substitution) will fail the completeness test if an “alt” is not provided in the instantiated components.
When you find yourself outgrowing the public UTA server, you may want to install your own UTA server. The instructions can be found on the [biocommons/uta repository README](https://bitbucket.org/biocommons/uta). Both “installing from database dumps” and installing from docker have been tested working with metavariant.
How to Set UTA Host Variables
metavariant connects to a UTA server as soon as it is imported, so reconfiguring which UTA server is used happens at the environment variable level. The relevant variables:
UTA_HOST (default: 'default') UTA_PORT (default: 5432) UTA_USER (default: 'uta_admin') UTA_PASS (default: 'uta_admin') UTA_SCHEMA (default: 'uta_20150903') UTA_TIMEOUT (default: 3) -- how long to wait before giving up on a connection
When you set up your own postgres server for the UTA database and you connect on the same server, the only environment variable you probably need to change is UTA_HOST (set it to “localhost”).
You may have to do more postgres administration to get your preferred configuration going, which is outside the scope of this README.
This library was developed in-house at Text2Gene, LLC.
It is provided to the community free of charge by way of the Apache 2.0 License.
You are free to modify it for commercial and non-commercial uses; just don’t try to sell it as-is.
Contributions, extensions, bug reports, suggestions, and swear words all happily accepted, in that order.
email@example.com 2016 and onwards