Plotting Tutorial

The Hail plot module allows for easy plotting of data. This notebook contains examples of how to use the plotting functions in this module, many of which can also be found in the first tutorial.

import hail as hl

from import show, output_notebook
from bokeh.layouts import gridplot
Running on Apache Spark version 2.4.1
SparkUI available at http://6d0e06cd602c:4040
Welcome to
     __  __     <>__
    / /_/ /__  __/ /
   / __  / _ `/ / /
  /_/ /_/\_,_/_/_/   version 0.2.61-3c86d3ba497a
LOGGING: writing to /oxj58bvohyvy/python/hail/docs/tutorials/hail-20201203-1906-0.2.61-3c86d3ba497a.log
Loading BokehJS ...
mt = hl.read_matrix_table('data/')
table = (hl.import_table('data/1kg_annotations.txt', impute=True)
mt = mt.annotate_cols(**table[mt.s])
mt = hl.sample_qc(mt)

2020-12-03 19:06:37 Hail: INFO: 1KG files found
2020-12-03 19:06:38 Hail: INFO: Reading table to impute column types
2020-12-03 19:06:41 Hail: INFO: Finished type imputation
  Loading field 'Sample' as type str (imputed)
  Loading field 'Population' as type str (imputed)
  Loading field 'SuperPopulation' as type str (imputed)
  Loading field 'isFemale' as type bool (imputed)
  Loading field 'PurpleHair' as type bool (imputed)
  Loading field 'CaffeineConsumption' as type int32 (imputed)
Global fields:
Column fields:
    's': str
    'Population': str
    'SuperPopulation': str
    'isFemale': bool
    'PurpleHair': bool
    'CaffeineConsumption': int32
    'sample_qc': struct {
        dp_stats: struct {
            mean: float64,
            stdev: float64,
            min: float64,
            max: float64
        gq_stats: struct {
            mean: float64,
            stdev: float64,
            min: float64,
            max: float64
        call_rate: float64,
        n_called: int64,
        n_not_called: int64,
        n_filtered: int64,
        n_hom_ref: int64,
        n_het: int64,
        n_hom_var: int64,
        n_non_ref: int64,
        n_singleton: int64,
        n_snp: int64,
        n_insertion: int64,
        n_deletion: int64,
        n_transition: int64,
        n_transversion: int64,
        n_star: int64,
        r_ti_tv: float64,
        r_het_hom_var: float64,
        r_insertion_deletion: float64
Row fields:
    'locus': locus<GRCh37>
    'alleles': array<str>
    'rsid': str
    'qual': float64
    'filters': set<str>
    'info': struct {
        AC: array<int32>,
        AF: array<float64>,
        AN: int32,
        BaseQRankSum: float64,
        ClippingRankSum: float64,
        DP: int32,
        DS: bool,
        FS: float64,
        HaplotypeScore: float64,
        InbreedingCoeff: float64,
        MLEAC: array<int32>,
        MLEAF: array<float64>,
        MQ: float64,
        MQ0: int32,
        MQRankSum: float64,
        QD: float64,
        ReadPosRankSum: float64,
        set: str
Entry fields:
    'GT': call
    'AD': array<int32>
    'DP': int32
    'GQ': int32
    'PL': array<int32>
Column key: ['s']
Row key: ['locus', 'alleles']


The histogram() method takes as an argument an aggregated hist expression, as well as optional arguments for the legend and title of the plot.

dp_hist = mt.aggregate_entries(hl.expr.aggregators.hist(mt.DP, 0, 30, 30))
p = hl.plot.histogram(dp_hist, legend='DP', title='DP Histogram')

This method, like all Hail plotting methods, also allows us to pass in fields of our data set directly. Choosing not to specify the range and bins arguments would result in a range being computed based on the largest and smallest values in the dataset and a default bins value of 50.

p = hl.plot.histogram(mt.DP, range=(0, 30), bins=30)

Cumulative Histogram

The cumulative_histogram() method works in a similar way to histogram().

p = hl.plot.cumulative_histogram(mt.DP, range=(0,30), bins=30)


The scatter() method can also take in either Python types or Hail fields as arguments for x and y.

p = hl.plot.scatter(mt.sample_qc.dp_stats.mean, mt.sample_qc.call_rate, xlabel='Mean DP', ylabel='Call Rate')

We can also pass in a Hail field as a label argument, which determines how to color the data points.

mt = mt.filter_cols((mt.sample_qc.dp_stats.mean >= 4) & (mt.sample_qc.call_rate >= 0.97))
ab = mt.AD[1] / hl.sum(mt.AD)
filter_condition_ab = ((mt.GT.is_hom_ref() & (ab <= 0.1)) |
                        (mt.GT.is_het() & (ab >= 0.25) & (ab <= 0.75)) |
                        (mt.GT.is_hom_var() & (ab >= 0.9)))
mt = mt.filter_entries(filter_condition_ab)
mt = hl.variant_qc(mt).cache()
common_mt = mt.filter_rows(mt.variant_qc.AF[1] > 0.01)
gwas = hl.linear_regression_rows(y=common_mt.CaffeineConsumption, x=common_mt.GT.n_alt_alleles(), covariates=[1.0])
pca_eigenvalues, pca_scores, _ = hl.hwe_normalized_pca(common_mt.GT)
2020-12-03 19:06:57 Hail: INFO: linear_regression_rows: running on 250 samples for 1 response variable y,
    with input variable x, and 1 additional covariate...
2020-12-03 19:07:01 Hail: INFO: hwe_normalized_pca: running PCA using 9087 variants.
2020-12-03 19:07:02 Hail: INFO: pca: running PCA with 10 components...
p = hl.plot.scatter(pca_scores.scores[0], pca_scores.scores[1],
                    title='PCA', xlabel='PC1', ylabel='PC2', collect_all=True)
2020-12-03 19:07:08 Hail: WARN: cols(): Resulting column table is sorted by 'col_key'.
    To preserve matrix table column order, first unkey columns with 'key_cols_by()'
2020-12-03 19:07:09 Hail: INFO: Coerced sorted dataset
2020-12-03 19:07:09 Hail: INFO: Coerced sorted dataset

Hail’s downsample aggregator is incorporated into the scatter(), qq(), and manhattan() functions. The collect_all parameter tells the plot function whether to collect all values or downsample. Choosing not to set this parameter results in downsampling.

p2 = hl.plot.scatter(pca_scores.scores[0], pca_scores.scores[1],
                    title='PCA (downsampled)', xlabel='PC1', ylabel='PC2', collect_all=False, n_divisions=50)

show(gridplot([p, p2], ncols=2, plot_width=400, plot_height=400))
2020-12-03 19:07:10 Hail: INFO: Coerced sorted dataset
2020-12-03 19:07:10 Hail: INFO: Coerced sorted dataset

2-D histogram

For visualizing relationships between variables in large datasets (where scatter plots may be less informative since they highlight outliers), the histogram_2d() function will create a heatmap with the number of observations in each section of a 2-d grid based on two variables.

p = hl.plot.histogram2d(pca_scores.scores[0], pca_scores.scores[1])
2020-12-03 19:07:10 Hail: WARN: At least one range was not defined in histogram_2d. Doing two passes...
2020-12-03 19:07:12 Hail: INFO: Ordering unsorted dataset with network shuffle

Q-Q (Quantile-Quantile)

The qq() function requires either a Python type or a Hail field containing p-values to be plotted. This function also allows for downsampling.

p = hl.plot.qq(gwas.p_value, collect_all=True)
p2 = hl.plot.qq(gwas.p_value, n_divisions=75)

show(gridplot([p, p2], ncols=2, plot_width=400, plot_height=400))
2020-12-03 19:07:17 Hail: INFO: Ordering unsorted dataset with network shuffle
2020-12-03 19:07:20 Hail: INFO: Ordering unsorted dataset with network shuffle


The manhattan() function requires a Hail field containing p-values.

p = hl.plot.manhattan(gwas.p_value)

We can also pass in a dictionary of fields that we would like to show up as we hover over a data point, and choose not to downsample if the dataset is relatively small.

hover_fields = dict([('alleles', gwas.alleles)])
p = hl.plot.manhattan(gwas.p_value, hover_fields=hover_fields, collect_all=True)