Quick Start in Python

We here show how to run geno-wide analysis using plink and bgen files using geno-sugar.

Bgen file example

import numpy as np
from numpy.random import RandomState
from bgen_reader import read_bgen, compute_dosage, allele_expectation
from sklearn.impute import SimpleImputer

import geno_sugar as gs
import geno_sugar.preprocess as prep

gs.download(
    "https://github.com/limix/bgen-reader-py/blob/master/example/example.bgen?raw=true"
)

# define random state
random = RandomState(1)

# import genotype file
bgen_file = "example.bgen"
bgen = read_bgen(bgen_file, verbose=False)
bim = bgen["variants"]
bim["i"] = range(len(bim))
G = compute_dosage(allele_expectation(bgen["genotype"], nalleles=2, ploidy=2))
print(bim)
print(G)

# subsample snps
Isnp = gs.is_in(bim, ("01", 10000, 14000))
G, bim = gs.snp_query(G, bim, Isnp)

# define geno preprocessing function
imputer = SimpleImputer(missing_values=np.nan, strategy="mean")
preprocess = prep.compose(
    [
        prep.filter_by_missing(max_miss=0.10),
        prep.impute(imputer),
        prep.filter_by_maf(min_maf=0.10),
        prep.standardize(),
    ]
)

# loop on geno
queue = gs.GenoQueue(G, bim, batch_size=200, preprocess=preprocess)
for _G, _bim in queue:
    # run genetic analysis
    print("Result:", np.einsum("is,is->", random.rand(*_G.shape), _G))

            id      rsid chrom     pos  nalleles allele_ids    i
0      SNPID_2    RSID_2    01    2000         2        A,G    0
1      SNPID_3    RSID_3    01    3000         2        A,G    1
2      SNPID_4    RSID_4    01    4000         2        A,G    2
3      SNPID_5    RSID_5    01    5000         2        A,G    3
4      SNPID_6    RSID_6    01    6000         2        A,G    4
5      SNPID_7    RSID_7    01    7000         2        A,G    5
6      SNPID_8    RSID_8    01    8000         2        A,G    6
7      SNPID_9    RSID_9    01    9000         2        A,G    7
8     SNPID_10   RSID_10    01   10000         2        A,G    8
9     SNPID_11   RSID_11    01   11000         2        A,G    9
10    SNPID_12   RSID_12    01   12000         2        A,G   10
11    SNPID_13   RSID_13    01   13000         2        A,G   11
12    SNPID_14   RSID_14    01   14000         2        A,G   12
13    SNPID_15   RSID_15    01   15000         2        A,G   13
14    SNPID_16   RSID_16    01   16000         2        A,G   14
15    SNPID_17   RSID_17    01   17000         2        A,G   15
16    SNPID_18   RSID_18    01   18000         2        A,G   16
17    SNPID_19   RSID_19    01   19000         2        A,G   17
18    SNPID_20   RSID_20    01   20000         2        A,G   18
19    SNPID_21   RSID_21    01   21000         2        A,G   19
20    SNPID_22   RSID_22    01   22000         2        A,G   20
21    SNPID_23   RSID_23    01   23000         2        A,G   21
22    SNPID_24   RSID_24    01   24000         2        A,G   22
23    SNPID_25   RSID_25    01   25000         2        A,G   23
24    SNPID_26   RSID_26    01   26000         2        A,G   24
25    SNPID_27   RSID_27    01   27000         2        A,G   25
26    SNPID_28   RSID_28    01   28000         2        A,G   26
27    SNPID_29   RSID_29    01   29000         2        A,G   27
28    SNPID_30   RSID_30    01   30000         2        A,G   28
29    SNPID_31   RSID_31    01   31000         2        A,G   29
..         ...       ...   ...     ...       ...        ...  ...
169  SNPID_171  RSID_171    01   71001         2        A,G  169
170  SNPID_172  RSID_172    01   72001         2        A,G  170
171  SNPID_173  RSID_173    01   73001         2        A,G  171
172  SNPID_174  RSID_174    01   74001         2        A,G  172
173  SNPID_175  RSID_175    01   75001         2        A,G  173
174  SNPID_176  RSID_176    01   76001         2        A,G  174
175  SNPID_177  RSID_177    01   77001         2        A,G  175
176  SNPID_178  RSID_178    01   78001         2        A,G  176
177  SNPID_179  RSID_179    01   79001         2        A,G  177
178  SNPID_180  RSID_180    01   80001         2        A,G  178
179  SNPID_181  RSID_181    01   81001         2        A,G  179
180  SNPID_182  RSID_182    01   82001         2        A,G  180
181  SNPID_183  RSID_183    01   83001         2        A,G  181
182  SNPID_184  RSID_184    01   84001         2        A,G  182
183  SNPID_185  RSID_185    01   85001         2        A,G  183
184  SNPID_186  RSID_186    01   86001         2        A,G  184
185  SNPID_187  RSID_187    01   87001         2        A,G  185
186  SNPID_188  RSID_188    01   88001         2        A,G  186
187  SNPID_189  RSID_189    01   89001         2        A,G  187
188  SNPID_190  RSID_190    01   90001         2        A,G  188
189  SNPID_191  RSID_191    01   91001         2        A,G  189
190  SNPID_192  RSID_192    01   92001         2        A,G  190
191  SNPID_193  RSID_193    01   93001         2        A,G  191
192  SNPID_194  RSID_194    01   94001         2        A,G  192
193  SNPID_195  RSID_195    01   95001         2        A,G  193
194  SNPID_196  RSID_196    01   96001         2        A,G  194
195  SNPID_197  RSID_197    01   97001         2        A,G  195
196  SNPID_198  RSID_198    01   98001         2        A,G  196
197  SNPID_199  RSID_199    01   99001         2        A,G  197
198  SNPID_200  RSID_200    01  100001         2        A,G  198
<BLANKLINE>
[199 rows x 7 columns]
dask.array<getitem, shape=(199, 500), dtype=float64, chunksize=(199, 500)>
.. read 9 / 9 variants (100.00%)
Result: -13.85549311486657