|
| 1 | +import unittest |
| 2 | + |
| 3 | +from v03_pipeline.lib.core.environment import Env |
| 4 | +from v03_pipeline.lib.misc.clickhouse import get_clickhouse_client |
| 5 | +from v03_pipeline.ops.repartition_clickhouse_grch38_snv_indel import ( |
| 6 | + REPARTITION_DATABASE_NAME, |
| 7 | + main, |
| 8 | +) |
| 9 | + |
| 10 | + |
| 11 | +class RepartitionGRCh38SnvIndelTest(unittest.TestCase): |
| 12 | + def setUp(self): |
| 13 | + client = get_clickhouse_client() |
| 14 | + client.execute( |
| 15 | + f""" |
| 16 | + DROP DATABASE IF EXISTS {Env.CLICKHOUSE_DATABASE} |
| 17 | + PARALLEL WITH |
| 18 | + DROP DATABASE IF EXISTS {REPARTITION_DATABASE_NAME}; |
| 19 | + """, |
| 20 | + ) |
| 21 | + client.execute( |
| 22 | + f""" |
| 23 | + CREATE DATABASE {Env.CLICKHOUSE_DATABASE} |
| 24 | + PARALLEL WITH |
| 25 | + CREATE DATABASE {REPARTITION_DATABASE_NAME}; |
| 26 | + """, |
| 27 | + ) |
| 28 | + client.execute( |
| 29 | + f""" |
| 30 | + CREATE TABLE {Env.CLICKHOUSE_DATABASE}.`GRCh38/SNV_INDEL/entries` ( |
| 31 | + `key` UInt32, |
| 32 | + `project_guid` LowCardinality(String), |
| 33 | + `family_guid` String, |
| 34 | + `is_annotated_in_any_gene` Boolean, |
| 35 | + `sign` Int8, |
| 36 | + `n_partitions` UInt8 MATERIALIZED 2, |
| 37 | + `partition_id` UInt8 MATERIALIZED farmHash64(family_guid) % n_partitions, |
| 38 | + PROJECTION xpos_projection |
| 39 | + ( |
| 40 | + SELECT * |
| 41 | + ORDER BY is_annotated_in_any_gene |
| 42 | + ) |
| 43 | + ) |
| 44 | + ENGINE = CollapsingMergeTree(sign) |
| 45 | + PARTITION BY project_guid |
| 46 | + ORDER BY (project_guid, family_guid, key) |
| 47 | + SETTINGS deduplicate_merge_projection_mode = 'rebuild'; |
| 48 | + """, |
| 49 | + ) |
| 50 | + client.execute( |
| 51 | + f""" |
| 52 | + INSERT INTO {Env.CLICKHOUSE_DATABASE}.`GRCh38/SNV_INDEL/entries` |
| 53 | + VALUES |
| 54 | + (0, 'project_a', 'family_a1', 0, 1), |
| 55 | + (1, 'project_a', 'family_a2', 0, 1), |
| 56 | + (2, 'project_a', 'family_a3', 0, 1), |
| 57 | + (0, 'project_b', 'family_b1', 0, 1), |
| 58 | + (1, 'project_b', 'family_b2', 0, 1), |
| 59 | + (2, 'project_b', 'family_b2', 0, 1), |
| 60 | + (0, 'project_c', 'family_c1', 1, 1), |
| 61 | + (3, 'project_c', 'family_c2', 1, 1), |
| 62 | + """, |
| 63 | + ) |
| 64 | + client.execute( |
| 65 | + f""" |
| 66 | + CREATE DICTIONARY {Env.CLICKHOUSE_DATABASE}.`GRCh38/SNV_INDEL/project_partitions_dict` |
| 67 | + ( |
| 68 | + `project_guid` String, |
| 69 | + `n_partitions` UInt8 |
| 70 | + ) |
| 71 | + PRIMARY KEY project_guid |
| 72 | + SOURCE( |
| 73 | + CLICKHOUSE( |
| 74 | + USER '{Env.CLICKHOUSE_WRITER_USER}' PASSWORD '{Env.CLICKHOUSE_WRITER_PASSWORD}' |
| 75 | + DB {Env.CLICKHOUSE_DATABASE} QUERY 'SELECT project_guid, 3 FROM `GRCh38/SNV_INDEL/entries`' |
| 76 | + ) |
| 77 | + ) |
| 78 | + LIFETIME(0) |
| 79 | + LAYOUT(FLAT(MAX_ARRAY_SIZE 10000)) |
| 80 | + """, |
| 81 | + ) |
| 82 | + |
| 83 | + def test_main_all_projects(self): |
| 84 | + client = get_clickhouse_client() |
| 85 | + main(1, []) |
| 86 | + res = client.execute( |
| 87 | + f""" |
| 88 | + SELECT *, n_partitions, partition_id FROM {REPARTITION_DATABASE_NAME}.`GRCh38/SNV_INDEL/repartitioned_entries` |
| 89 | + """, |
| 90 | + ) |
| 91 | + self.assertCountEqual( |
| 92 | + res, |
| 93 | + [ |
| 94 | + (3, 'project_c', 'family_c2', True, 1, 2, 1), |
| 95 | + (0, 'project_b', 'family_b1', 0, 1, 2, 1), |
| 96 | + (1, 'project_b', 'family_b2', 0, 1, 2, 1), |
| 97 | + (2, 'project_b', 'family_b2', 0, 1, 2, 1), |
| 98 | + (2, 'project_a', 'family_a3', 0, 1, 2, 0), |
| 99 | + (0, 'project_a', 'family_a1', 0, 1, 2, 1), |
| 100 | + (1, 'project_a', 'family_a2', 0, 1, 2, 1), |
| 101 | + (0, 'project_c', 'family_c1', True, 1, 2, 0), |
| 102 | + ], |
| 103 | + ) |
| 104 | + |
| 105 | + def test_main_one_project(self): |
| 106 | + client = get_clickhouse_client() |
| 107 | + main(1, ['project_a']) |
| 108 | + res = client.execute( |
| 109 | + f""" |
| 110 | + SELECT *, n_partitions, partition_id FROM {REPARTITION_DATABASE_NAME}.`GRCh38/SNV_INDEL/repartitioned_entries` |
| 111 | + """, |
| 112 | + ) |
| 113 | + self.assertCountEqual( |
| 114 | + res, |
| 115 | + [ |
| 116 | + (2, 'project_a', 'family_a3', 0, 1, 2, 0), |
| 117 | + (0, 'project_a', 'family_a1', 0, 1, 2, 1), |
| 118 | + (1, 'project_a', 'family_a2', 0, 1, 2, 1), |
| 119 | + ], |
| 120 | + ) |
0 commit comments