Skip to main content

a lexical manipulation toolkit for genetic variant descriptors (hgvs, etc)

Project description

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:


The above “should” be successfully installed by using and/or by installing this package via pip.

Variant: shortcut to SequenceVariant

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 for explanation of HGVS strings.

Examples of valid HGVS strings:


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: Lexical Variant Generation

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.

Usage example:

hgvs_text = 'NM_198056.2:c.4786T>A'
lex = VariantLVG(hgvs_text)



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.

The done-for-you way of enriching the LVG is by using the NCBIEnrichedLVG object.

Usage example:

hgvs_text = 'NM_198056.2:c.4786T>A'
lex = NCBIEnrichedLVG(hgvs_text)

This object takes the first step of asking the NCBI Variant Reporter about the given HGVS string to see if the NCBI has any human-curated equivalents already. If so, NCBIEnrichedLVG uses these variant hgvs strings to feed into the VariantLVG process.

The result is an LVG object just like VariantLVG, except it probably took an extra 3-5 seconds to get results. (Hey, the NCBI is federally funded; we just pass that savings down to you.) The upside is that you might have a more well-featured LVG object that covers more lexical variations than you would have otherwise. Maybe.

Note: this method will be changing radically in the next minor version (0.3.0) since the NCBI Variant Reporter has already been sunsetted in favor of Variation Services. See `the NCBI Variation Services page <>`_.

Keyword Arguments

  • transcripts (list): list of strings describing valid alternative transcripts for seqvar

  • seqvar_max_len (int): restrict posedit lengths to this number of characters (or fewer).

  • hgvs_c (list): see Enrichment above

  • hgvs_g (list): see Enrichment above

  • hgvs_n (list): see Enrichment above

  • hgvs_p (list): see Enrichment above


  • hgvs_text: original hgvs string from instantiation

  • seqvar: original SequenceVariant from instantiation

  • transcripts: list of strings indicating related transcripts

  • variants: 2-level dictionary of shape { seqtype: { hgvs_text: seqvar } }


  • gene_name: returns HUGO gene-name if it can be ascertained using UTA. (Lazy-loaded attribute _gene_name.)

  • hgvs_c: returns flat list of c.DNA hgvs strings from variants

  • hgvs_g: returns flat list of g.DNA hgvs strings from variants

  • hgvs_n: returns flat list of n.RNA hgvs strings from variants

  • hgvs_p: returns flat list of protein hgvs strings from variants

  • seqvars: returns flat list of SequenceVariant objects from variants


  • to_dict(): returns non-underscored attributes (seqvar, hgvs_text, transcripts, seqvars) as dictionary

  • to_json(): returns a serialized JSON string representation of the object which can be used to instantiate this LVG again.

  • from_json(json_str): takes serialized JSON representation of this object and rebuilds LVG from its details.

VariantComponents: Parsing and “Slang”

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')

Usage starting from “aminochange” string:

comp = VariantComponents(aminochange='V777A')
comp = VariantComponents(aminochange='Leu653Gly')

(If starting with “aminochange” string, the seqvar and edittype attributes will be None.)

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.

MetaVariantException: base exception class from which all metavariant package exceptions are subclassed.

Setting UTA Server

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]( 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.

Support and Maintenance

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. 2016 and onwards

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

metavariant-0.2.0.tar.gz (27.1 kB view hashes)

Uploaded Source

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page