diff options
Diffstat (limited to 'gn3/db')
| -rw-r--r-- | gn3/db/correlations.py | 109 | ||||
| -rw-r--r-- | gn3/db/species.py | 20 |
2 files changed, 94 insertions, 35 deletions
diff --git a/gn3/db/correlations.py b/gn3/db/correlations.py index 87ab082..06b3310 100644 --- a/gn3/db/correlations.py +++ b/gn3/db/correlations.py @@ -10,6 +10,8 @@ from gn3.random import random_string from gn3.data_helpers import partition_all from gn3.db.species import translate_to_mouse_gene_id +from gn3.computations.partial_correlations import correlations_of_all_tissue_traits + def get_filename(target_db_name: str, conn: Any) -> str: """ Retrieve the name of the reference database file with which correlations are @@ -140,22 +142,6 @@ def fetch_literature_correlations( cursor.execute("DROP TEMPORARY TABLE %s", temp_table) return dict(results) -def compare_tissue_correlation_absolute_values(val1, val2): - """ - Comparison function for use when sorting tissue correlation values. - - This is a partial migration of the - `web.webqtl.correlation.CorrelationPage.getTempTissueCorrTable` function in - GeneNetwork1.""" - try: - if abs(val1) < abs(val2): - return 1 - if abs(val1) == abs(val2): - return 0 - return -1 - except TypeError: - return 0 - def fetch_symbol_value_pair_dict( symbol_list: Tuple[str, ...], data_id_dict: dict, conn: Any) -> Dict[str, Tuple[float, ...]]: @@ -265,14 +251,21 @@ def fetch_tissue_probeset_xref_info( results or tuple(), (tuple(), {}, {}, {}, {}, {}, {})) -def correlations_of_all_tissue_traits() -> Tuple[dict, dict]: +def fetch_gene_symbol_tissue_value_dict_for_trait( + gene_name_list: Tuple[str, ...], probeset_freeze_id: int, + conn: Any) -> dict: """ + Fetches a map of the gene symbols to the tissue values. + This is a migration of the - `web.webqtl.correlation.CorrelationPage.calculateCorrOfAllTissueTrait` + `web.webqtl.correlation.correlationFunction.getGeneSymbolTissueValueDictForTrait` function in GeneNetwork1. """ - raise Exception("Unimplemented!!!") - return ({}, {}) + xref_info = fetch_tissue_probeset_xref_info( + gene_name_list, probeset_freeze_id, conn) + if xref_info[0]: + return fetch_gene_symbol_tissue_value_dict(xref_info[0], xref_info[2], conn) + return {} def build_temporary_tissue_correlations_table( trait_symbol: str, probeset_freeze_id: int, method: str, @@ -283,10 +276,40 @@ def build_temporary_tissue_correlations_table( This is a migration of the `web.webqtl.correlation.CorrelationPage.getTempTissueCorrTable` function in GeneNetwork1.""" - raise Exception("Unimplemented!!!") - return "" + # We should probably pass the `correlations_of_all_tissue_traits` function + # as an argument to this function and get rid of the one call immediately + # following this comment. + symbol_corr_dict, symbol_p_value_dict = correlations_of_all_tissue_traits( + fetch_gene_symbol_tissue_value_dict_for_trait( + (trait_symbol,), probeset_freeze_id, conn), + fetch_gene_symbol_tissue_value_dict_for_trait( + tuple(), probeset_freeze_id, conn), + method) + + symbol_corr_list = sorted( + symbol_corr_dict.items(), key=lambda key_val: key_val[1]) + + temp_table_name = f"TOPTISSUE{random_string(8)}" + create_query = ( + "CREATE TEMPORARY TABLE {temp_table_name}" + "(Symbol varchar(100) PRIMARY KEY, Correlation float, PValue float)") + insert_query = ( + f"INSERT INTO {temp_table_name}(Symbol, Correlation, PValue) " + " VALUES (%(symbol)s, %(correlation)s, %(pvalue)s)") -def fetch_tissue_correlations( + with conn.cursor() as cursor: + cursor.execute(create_query) + cursor.execute( + insert_query, + tuple({ + "symbol": symbol, + "correlation": corr, + "pvalue": symbol_p_value_dict[symbol] + } for symbol, corr in symbol_corr_list[0: 2 * return_number])) + + return temp_table_name + +def fetch_tissue_correlations(# pylint: disable=R0913 dataset: dict, trait_symbol: str, probeset_freeze_id: int, method: str, return_number: int, conn: Any) -> dict: """ @@ -316,3 +339,43 @@ def fetch_tissue_correlations( return { trait_name: (tiss_corr, tiss_p_val) for trait_name, tiss_corr, tiss_p_val in results} + +def check_for_literature_info(conn: Any, geneid: int) -> bool: + """ + Checks the database to find out whether the trait with `geneid` has any + associated literature. + + This is a migration of the + `web.webqtl.correlation.CorrelationPage.checkForLitInfo` function in + GeneNetwork1. + """ + query = "SELECT 1 FROM LCorrRamin3 WHERE GeneId1=%s LIMIT 1" + with conn.cursor() as cursor: + cursor.execute(query, geneid) + result = cursor.fetchone() + if result: + return True + + return False + +def check_symbol_for_tissue_correlation( + conn: Any, tissue_probeset_freeze_id: int, symbol: str = "") -> bool: + """ + Checks whether a symbol has any associated tissue correlations. + + This is a migration of the + `web.webqtl.correlation.CorrelationPage.checkSymbolForTissueCorr` function + in GeneNetwork1. + """ + query = ( + "SELECT 1 FROM TissueProbeSetXRef " + "WHERE TissueProbeSetFreezeId=%(probeset_freeze_id)s " + "AND Symbol=%(symbol)s LIMIT 1") + with conn.cursor() as cursor: + cursor.execute( + query, probeset_freeze_id=tissue_probeset_freeze_id, symbol=symbol) + result = cursor.fetchone() + if result: + return True + + return False diff --git a/gn3/db/species.py b/gn3/db/species.py index 1e5015f..702a9a8 100644 --- a/gn3/db/species.py +++ b/gn3/db/species.py @@ -47,17 +47,13 @@ def translate_to_mouse_gene_id(species: str, geneid: int, conn: Any) -> int: return geneid with conn.cursor as cursor: - if species == "rat": - cursor.execute( - "SELECT mouse FROM GeneIDXRef WHERE rat = %s", geneid) - rat_geneid = cursor.fetchone() - if rat_geneid: - return rat_geneid[0] - - cursor.execute( - "SELECT mouse FROM GeneIDXRef WHERE human = %s", geneid) - human_geneid = cursor.fetchone() - if human_geneid: - return human_geneid[0] + query = { + "rat": "SELECT mouse FROM GeneIDXRef WHERE rat = %s", + "human": "SELECT mouse FROM GeneIDXRef WHERE human = %s" + } + cursor.execute(query[species], geneid) + translated_gene_id = cursor.fetchone() + if translated_gene_id: + return translated_gene_id[0] return 0 # default if all else fails |