hgvs - manipulate biological sequence variants according to Human Genome Variation Society recommendations
Important: biocommons packages (including hgvs) will begin phasing out support for Python 2.7 on Jan 1, 2019. More
The hgvs package provides a Python library to parse, format, validate, normalize, and map sequence variants according to Variation Nomenclature (aka Human Genome Variation Society) recommendations.
| Information | 
|
| Latest Release | 
|
| Development | 
  
|
- Parsing is based on formal grammar.
- An easy-to-use object model that represents most variant types (SNVs, indels, dups, inverstions, etc) and concepts (intronic offsets, uncertain positions, intervals)
- A variant normalizer that rewrites variants in canoncial forms and substitutes reference sequences (if reference and transcript sequences differ)
- Formatters that generate HGVS strings from internal representations
- Tools to map variants between genome, transcript, and protein sequences
- Reliable handling of regions genome-transcript discrepancies
- Pluggable data providers support alternative sources of transcript mapping data
- Extensive automated tests, including those for all variant types and "problematic" transcripts
- Easily installed using remote data sources. Installation with local data sources is straightforward and completely obviates network access
- You are encouraged to browse issues. All known issues are listed there. Please report any issues you find.
- Use a pip package specification to stay within minor releases.
For example,
hgvs>=1.1,<1.2. hgvs uses Semantic Versioning.
By default, hgvs uses remote data sources, which makes installation easy.
$ mkvirtualenv hgvs-test (hgvs-test)$ pip install --upgrade setuptools (hgvs-test)$ pip install hgvs (hgvs-test)$ python
See Installation instructions for details, including instructions for installing Universal Transcript Archive (UTA) and SeqRepo locally.
hgvs parses HGVS variants (as strings) into an object model, and can format object models back into HGVS strings.
>>> import hgvs.parser # start with these variants as strings >>> hgvs_g = 'NC_000007.13:g.36561662C>T' >>> hgvs_c = 'NM_001637.3:c.1582G>A' # parse the genomic variant into a Python structure >>> hp = hgvs.parser.Parser() >>> var_g = hp.parse_hgvs_variant(hgvs_g) >>> var_g SequenceVariant(ac=NC_000007.13, type=g, posedit=36561662C>T) # SequenceVariants are composed of structured objects, e.g., >>> var_g.posedit.pos.start SimplePosition(base=36561662, uncertain=False) # format by stringification >>> str(var_g) 'NC_000007.13:g.36561662C>T'
hgvs provides tools to project variants between genome, transcript, and protein sequences. Non-coding and intronic variants are supported. Alignment data come from the Universal Transcript Archive (UTA).
>>> import hgvs.dataproviders.uta >>> import hgvs.assemblymapper # initialize the mapper for GRCh37 with splign-based alignments >>> hdp = hgvs.dataproviders.uta.connect() >>> am = hgvs.assemblymapper.AssemblyMapper(hdp, ... assembly_name='GRCh37', alt_aln_method='splign', ... replace_reference=True) # identify transcripts that overlap this genomic variant >>> transcripts = am.relevant_transcripts(var_g) >>> sorted(transcripts) ['NM_001177506.1', 'NM_001177507.1', 'NM_001637.3'] # map genomic variant to one of these transcripts >>> var_c = am.g_to_c(var_g, 'NM_001637.3') >>> var_c SequenceVariant(ac=NM_001637.3, type=c, posedit=1582G>A) >>> str(var_c) 'NM_001637.3:c.1582G>A' # CDS coordinates use BaseOffsetPosition to support intronic offsets >>> var_c.posedit.pos.start BaseOffsetPosition(base=1582, offset=0, datum=Datum.CDS_START, uncertain=False)
Coding variants may be translated to their protein consequences. HGVS uses the same pairing of transcript and protein accessions as seen in NCBI and Ensembl.
# translate var_c to its protein consequence
# The object structure of protein variants is nearly identical to
# that of nucleic acid variants and is converted to a string form
# by stringification. Per HGVS recommendations, inferred consequences
# must have parentheses to indicate uncertainty.
>>> var_p = am.c_to_p(var_c)
>>> var_p
SequenceVariant(ac=NP_001628.1, type=p, posedit=(Gly528Arg))
>>> str(var_p)
'NP_001628.1:p.(Gly528Arg)'
# setting uncertain to False removes the parentheses on the
# stringified form
>>> var_p.posedit.uncertain = False
>>> str(var_p)
'NP_001628.1:p.Gly528Arg'
# formatting can be customized, e.g., use 1 letter amino acids to
# format a specific variant
# (configuration may also be set globally)
>>> var_p.format(conf={"p_3_letter": False})
'NP_001628.1:p.G528R'
Some variants have multiple representations due to instrinsic biological ambiguity (e.g., inserting a G in a poly-G run) or due to misunderstanding HGVS recommendations. Normalization rewrites certain veriants into a single representation.
# rewrite ins as dup (depends on sequence context)
>>> import hgvs.normalizer
>>> hn = hgvs.normalizer.Normalizer(hdp)
>>> hn.normalize(hp.parse_hgvs_variant('NM_001166478.1:c.35_36insT'))
SequenceVariant(ac=NM_001166478.1, type=c, posedit=35dup)
# during mapping, variants are normalized (by default)
>>> c1 = hp.parse_hgvs_variant('NM_001166478.1:c.31del')
>>> c1
SequenceVariant(ac=NM_001166478.1, type=c, posedit=31del)
>>> c1n = hn.normalize(c1)
>>> c1n
SequenceVariant(ac=NM_001166478.1, type=c, posedit=35del)
>>> g = am.c_to_g(c1)
>>> g
SequenceVariant(ac=NC_000006.11, type=g, posedit=49917127del)
>>> c2 = am.g_to_c(g, c1.ac)
>>> c2
SequenceVariant(ac=NM_001166478.1, type=c, posedit=35del)
There are more examples in the documentation.
The hgvs package is intended to be a community project. Please see Contributing to get started in submitting source code, tests, or documentation. Thanks for getting involved!
Other packages that manipulate HGVS variants: