package varspark.hail

This package contains variant spark integration with Hail.

from hail import *
import varspark.hail
hc = HailContext(sc)
vds = hc.import_vcf(...)
...
via = vds.importance_analysis("sa.pheno.label", n_trees = 1000)

module varspark.hail.extend

Created on 7 Nov 2017

@author: szu004

class varspark.hail.extend.VariantsDatasetFunctions(*args, **kwargs)

Extension to hail.VariantDataset with variant-spark related functions

importance_analysis(**kwargs)

Builds random forest classifier for the response variable defined with y_expr.

Parameters:

• y_expr (str) – Response expression. Must evaluate to Boolean or numeric with all values 0 or 1.
• n_trees (int) – The number of trees to build in the forest.
• mtry_fraction (float) – The fraction of variables to try at each split.
• oob (bool) – Should OOB error be calculated.
• seed (long) – Random seed to use.
• batch_size (int) – The number of trees to build in one batch.

Returns:

Importance analysis model.

Return type:

ImportanceAnalysis

pairwise_operation(**kwargs)

Computes a pairwise operation on encoded genotypes. Currently implemented operations include:

• manhattan : the Manhattan distance
• euclidean : the Euclidean distance
• sharedAltAlleleCount: count of shared alternative alleles
• anySharedAltAlleleCount: count of variants that share at least one alternative allele

Parameters:operation_name – name of the operaiton. One of manhattan, euclidean, sharedAltAlleleCount, anySharedAltAlleleCount Returns:A symmetric no_of_samples x no_of_samples matrix with the result of the pairwise computation. Return type:hail.KinshipMatrix

module varspark.hail.rf

Created on 10 Nov 2017

@author: szu004

class varspark.hail.rf.ImportanceAnalysis(hc, _jia)

Model for random forest based importance analysis

important_variants(**kwargs)

Gets the top n most important loci.

Parameters:n_limit (int) – the limit of the number of loci to return Returns:A KeyTable with the variant in the first column and importance in the second. Return type:hail.KeyTable

oob_error

OOB (Out of Bag) error estimate for the model

Return type:float