update sqlite database location

patflick · Sep 22, 2014 · 7066d4e · 7066d4e
1 parent 94dddc5
commit 7066d4e
Show file tree

Hide file tree

Showing 17 changed files with 39 additions and 38 deletions.
diff --git a/analysis/bossi_lehner.R b/analysis/bossi_lehner.R
@@ -48,7 +48,7 @@ get_expression_property_data <- function(ppi_name="bossi",expr_name="gene_atlas"
 {
  # load sql config and get connection
  source("sql_config.R")
- con <- get_sql_conn('/home/patrick/dev/bio/data/test_matching.sqlite')
+ con <- get_sql_conn()
 
  # create table names
  prop_table <- paste(ppi_name, expr_name, "agg_node_properties", sep="_")
@@ -109,7 +109,7 @@ plot_tissue_expr_count_hist <- function(expr_name="gene_atlas")
 {
  # load the ts/hk summary data from the database
  source("sql_config.R")
- con <- get_sql_conn('/home/patrick/dev/bio/data/test_matching.sqlite')
+ con <- get_sql_conn()
 
  expr_count_table <- paste(expr_name, "core_expr_counts", sep="_")
  #expr_count_table <- paste(expr_name, "expr_counts", sep="_")
@@ -207,7 +207,7 @@ get_min_max_neighbor_expr_data <- function(ppi_name="bossi", expr_name="gene_atl
 
  # load the ts/hk summary data from the database
  source("sql_config.R")
- con <- get_sql_conn('/home/patrick/dev/bio/data/test_matching.sqlite')
+ con <- get_sql_conn()
 
  # have to load from `core`, as the min and max counts are also
  # calculated in the `core` of the expression data set.
@@ -594,7 +594,7 @@ get_hkts_edge_summary <- function(ppi_name="bossi", expr_name="gene_atlas", prop
  # tissue specific proteins by only using SQL:
 
  source("sql_config.R")
- con <- get_sql_conn('/home/patrick/dev/bio/data/test_matching.sqlite')
+ con <- get_sql_conn()
 
  # what's the definition of tissuespecific vs. housekeeping?
  hkts_table <- get_hkts_sym_threshold_table(expr_name, 0.2)
@@ -616,7 +616,7 @@ create_test_degrees <- function(ppi_name="bossi",expr_name="gene_atlas",prop_nam
 {
 
  source("sql_config.R")
- con <- get_sql_conn('/home/patrick/dev/bio/data/test_matching.sqlite')
+ con <- get_sql_conn()
 
  # create table names
  prop_table <- paste("prop", ppi_name, expr_name, prop_name, sep="_")
@@ -639,7 +639,7 @@ compare_test_degrees <- function(ppi_name="bossi",expr_name="gene_atlas",prop_na
 {
 
  source("sql_config.R")
- con <- get_sql_conn('/home/patrick/dev/bio/data/test_matching.sqlite')
+ con <- get_sql_conn()
 
  # create table names
  prop_table_1 <- paste("prop", ppi_name, expr_name, "degree", sep="_")

diff --git a/analysis/cluster_score_hist.R b/analysis/cluster_score_hist.R
@@ -18,7 +18,7 @@ clusterers <- c('PLP','PLM-gamma-1.0', 'PLM-gamma-5.0', 'PLM-gamma-10.0', 'PLM-g
 get_cluster_data <- function(ppi_name="string", expr_name="gene_atlas", clusterer=clusterers[1])
 {
  source("sql_config.R")
- con <- get_sql_conn('/home/patrick/dev/bio/data/test_matching.sqlite')
+ con <- get_sql_conn()
  query <- paste("SELECT * FROM clustering_scoring_results ",
  "WHERE size >= 4 AND ppi='", ppi_name,
  "' AND expr='",expr_name, "'",
@@ -31,7 +31,7 @@ get_clusterer_data <- function(ppi_name="string", expr_name="gene_atlas")
 {
  type <- "Global"
  source("sql_config.R")
- con <- get_sql_conn('/home/patrick/dev/bio/data/test_matching.sqlite')
+ con <- get_sql_conn()
  query <- paste("SELECT ppi, expr, clusterer, SUM(modularity) as total_mod, COUNT(), MIN(size), AVG(size), MAX(size) FROM clustering_scoring_results ",
  "WHERE",
  " size >= 4",
@@ -48,7 +48,7 @@ plot_cluster_distr <- function(ppi_name="string", expr_name="gene_atlas")
 {
  type <- "GLOBAL"
  source("sql_config.R")
- con <- get_sql_conn('/home/patrick/dev/bio/data/test_matching.sqlite')
+ con <- get_sql_conn()
  query <- paste("SELECT * FROM clustering_scoring_results ",
  "WHERE ppi='", ppi_name,
  "' AND expr='",expr_name, "'",
@@ -85,7 +85,7 @@ get_top_cluster_distr <- function(ppi_name="string", expr_name="gene_atlas")
 {
  type <- "GLOBAL"
  source("sql_config.R")
- con <- get_sql_conn('/home/patrick/dev/bio/data/test_matching.sqlite')
+ con <- get_sql_conn()
  query <- paste("SELECT * FROM clustering_scoring_results ",
  "WHERE ppi='", ppi_name,
  "' AND expr='",expr_name, "'",

diff --git a/analysis/expr_core.R b/analysis/expr_core.R
@@ -13,7 +13,7 @@ is_core <- function(expr_name="hpa")
 {
  # load the ts/hk summary data from the database
  source("sql_config.R")
- con <- get_sql_conn('/home/patrick/dev/bio/data/test_matching.sqlite')
+ con <- get_sql_conn()
 
  query <- paste("SELECT COUNT(DISTINCT Gene) FROM ", expr_name)
  gene_count <- dbGetQuery(con, query)

diff --git a/analysis/expr_stats.R b/analysis/expr_stats.R
@@ -8,7 +8,7 @@ source("expr_utils.R")
 get_all_expr_size_stats <- function()
 {
  source("sql_config.R")
- con <- get_sql_conn('/home/patrick/dev/bio/data/test_matching.sqlite')
+ con <- get_sql_conn()
 
  for (e in get_exprs())
  {
@@ -27,7 +27,7 @@ get_expr_values <- function(expr_name="gene_atlas")
 {
  # load the ts/hk summary data from the database
  source("sql_config.R")
- con <- get_sql_conn('/home/patrick/dev/bio/data/test_matching.sqlite')
+ con <- get_sql_conn()
 
  expr_normalized_table <- paste(expr_name, "normalized", sep="_")
  query <- paste("SELECT * FROM ", expr_normalized_table)
@@ -61,7 +61,7 @@ plot_tissue_expr_count_hist <- function(expr_name="gene_atlas")
 {
  # load the ts/hk summary data from the database
  source("sql_config.R")
- con <- get_sql_conn('/home/patrick/dev/bio/data/test_matching.sqlite')
+ con <- get_sql_conn()
 
  expr_count_table <- paste(expr_name, "expr_counts", sep="_")
  query <- paste("SELECT * FROM ", expr_count_table, " ORDER BY ExpressedCount*1.0/TotalCount")
@@ -84,7 +84,7 @@ plot_tshk_example <- function(expr_name="hpa", threshold=0.15)
 {
  # load the ts/hk summary data from the database
  source("sql_config.R")
- con <- get_sql_conn('/home/patrick/dev/bio/data/test_matching.sqlite')
+ con <- get_sql_conn()
 
  expr_count_table <- paste(expr_name, "expr_counts", sep="_")
  query <- paste("SELECT * FROM ", expr_count_table, " WHERE ExpressedCount > 0 ORDER BY ExpressedCount*1.0/TotalCount")
@@ -129,7 +129,7 @@ plot_tshk_thresholds <- function(expr_name="hpa")
 {
  # load the ts/hk summary data from the database
  source("sql_config.R")
- con <- get_sql_conn('/home/patrick/dev/bio/data/test_matching.sqlite')
+ con <- get_sql_conn()
 
  expr_count_table <- paste(expr_name, "expr_counts", sep="_")
  query <- paste("SELECT * FROM ", expr_count_table, " WHERE ExpressedCount > 0 ORDER BY ExpressedCount*1.0/TotalCount")

diff --git a/analysis/incr_betweenness.R b/analysis/incr_betweenness.R
@@ -40,7 +40,7 @@ get_incr_query <- function(ppi_name="ccsb", expr_name="hpa")
 create_union_incr_table <- function()
 {
  source("sql_config.R")
- con <- get_sql_conn('/home/patrick/dev/bio/data/test_matching.sqlite')
+ con <- get_sql_conn()
 
  queries <- c()
  for (p in get_ppis())
@@ -64,7 +64,7 @@ create_union_incr_table <- function()
 get_incr_betweenness_nodes <- function()
 {
  source("sql_config.R")
- con <- get_sql_conn('/home/patrick/dev/bio/data/test_matching.sqlite')
+ con <- get_sql_conn()
 
  query <- paste("SELECT * FROM ", SQL_INCR_TABLE_NAME)
  data <- dbGetQuery(con, query)
@@ -81,7 +81,7 @@ plot_betw_incr_distr <- function()
 get_num_genes <- function()
 {
  source("sql_config.R")
- con <- get_sql_conn('/home/patrick/dev/bio/data/test_matching.sqlite')
+ con <- get_sql_conn()
 
  query <- paste("SELECT ppi, expr, COUNT(DISTINCT Gene) FROM ",
  SQL_INCR_TABLE_NAME,
@@ -95,7 +95,7 @@ get_num_genes <- function()
 get_genes_count <- function()
 {
  source("sql_config.R")
- con <- get_sql_conn('/home/patrick/dev/bio/data/test_matching.sqlite')
+ con <- get_sql_conn()
 
  query <- paste("SELECT Gene, COUNT(DISTINCT ppi) as cnt_ppi, ",
  " COUNT(DISTINCT expr) as cnt_expr, ",

diff --git a/analysis/old/overlap_analysis.R b/analysis/old/overlap_analysis.R
@@ -1,7 +1,7 @@
 
 # load sql config and get connection
 source("sql_config.R")
-con <- get_sql_conn('/home/patrick/dev/bio/data/test_matching.sqlite')
+con <- get_sql_conn()
 
 table <- "overlap_pairwise_ppi_ids"
 ppi_id_data <- dbGetQuery(con, paste("SELECT * FROM ", table))

diff --git a/analysis/overlap_plots.R b/analysis/overlap_plots.R
@@ -7,7 +7,7 @@ plot_pairwise_table <- function(src_table)
 {
  # load sql config and get connection
  source("sql_config.R")
- con <- get_sql_conn('/home/patrick/dev/bio/data/test_matching.sqlite')
+ con <- get_sql_conn()
 
  # load ppi network from db
  data <- dbGetQuery(con, paste("SELECT * FROM ", src_table))

diff --git a/analysis/ppi_stats.R b/analysis/ppi_stats.R
@@ -11,7 +11,7 @@ get_node_properties <- function(ppi_name = "string")
 {
  # load the ts/hk summary data from the database
  source("sql_config.R")
- con <- get_sql_conn('/home/patrick/dev/bio/data/test_matching.sqlite')
+ con <- get_sql_conn()
 
  ppi_node_prop_tbl <- paste(ppi_name, "node_properties", sep="_")
  query <- paste("SELECT * FROM ", ppi_node_prop_tbl)
@@ -24,7 +24,7 @@ get_graph_properties <- function(ppi_name="string")
 {
  # load the ts/hk summary data from the database
  source("sql_config.R")
- con <- get_sql_conn('/home/patrick/dev/bio/data/test_matching.sqlite')
+ con <- get_sql_conn()
 
  ppi_prop_tbl <- paste(ppi_name, "properties", sep="_")
  query <- paste("SELECT * FROM ", ppi_prop_tbl)

diff --git a/analysis/sql_config.R b/analysis/sql_config.R
@@ -2,7 +2,7 @@
 # 
 # Author: flick
 ###############################################################################
-DEFAULT_DATABASE_FILE = 'hpaDB.sqlite'
+DEFAULT_DATABASE_FILE = '../data/sql_database.sqlite'
 
 
 get_sql_conn <- function(database_file=DEFAULT_DATABASE_FILE)

diff --git a/analysis/ts_betweenness.R b/analysis/ts_betweenness.R
@@ -21,7 +21,7 @@ get_node_properties <- function(ppi_name = "string", expr_name = "gene_atlas")
 {
  # load the ts/hk summary data from the database
  source("sql_config.R")
- con <- get_sql_conn('/home/patrick/dev/bio/data/test_matching.sqlite')
+ con <- get_sql_conn()
 
  ppi_node_prop_tbl <- paste(ppi_name,expr_name, "node_properties", sep="_")
  query <- paste("SELECT * FROM ", ppi_node_prop_tbl)
@@ -36,7 +36,7 @@ get_ts_node_properties <- function(ppi_name = "string", expr_name = "gene_atlas"
 {
  # load the ts/hk summary data from the database
  source("sql_config.R")
- con <- get_sql_conn('/home/patrick/dev/bio/data/test_matching.sqlite')
+ con <- get_sql_conn()
 
  ppi_node_prop_tbl <- paste(ppi_name,expr_name, "ts_node_properties", sep="_")
  query <- paste("SELECT * FROM ", ppi_node_prop_tbl)
@@ -49,7 +49,7 @@ get_node_properties_exprcount <- function(ppi_name="string",expr_name="gene_atla
 {
  # load sql config and get connection
  source("sql_config.R")
- con <- get_sql_conn('/home/patrick/dev/bio/data/test_matching.sqlite')
+ con <- get_sql_conn()
 
  # create table names
  ppi_node_prop_tbl <- paste(ppi_name,expr_name, "node_properties", sep="_")
@@ -69,7 +69,7 @@ get_ts_node_properties_exprcount <- function(ppi_name="string",expr_name="gene_a
 {
  # load sql config and get connection
  source("sql_config.R")
- con <- get_sql_conn('/home/patrick/dev/bio/data/test_matching.sqlite')
+ con <- get_sql_conn()
 
  # create table names
  ppi_node_prop_tbl <- paste(ppi_name,expr_name, "ts_node_properties", sep="_")

diff --git a/analysis/tsppi_stats.R b/analysis/tsppi_stats.R
@@ -14,7 +14,7 @@ get_graph_properties <- function(ppi_name="string", expr_name="gene_atlas")
 {
  # load the ts/hk summary data from the database
  source("sql_config.R")
- con <- get_sql_conn('/home/patrick/dev/bio/data/test_matching.sqlite')
+ con <- get_sql_conn()
 
  ppi_prop_tbl <- paste(ppi_name, expr_name, "properties", sep="_")
  query <- paste("SELECT * FROM ", ppi_prop_tbl)
@@ -27,7 +27,7 @@ get_ts_graph_properties <- function()
 {
  # load the ts/hk summary data from the database
  source("sql_config.R")
- con <- get_sql_conn('/home/patrick/dev/bio/data/test_matching.sqlite')
+ con <- get_sql_conn()
 
  ppi_prop_tbl <- "ts_graph_properties"
  query <- paste("SELECT ppi, expr, Property, AVG(Value) as avg_value FROM ",

diff --git a/analysis/venn_diagram.R b/analysis/venn_diagram.R
@@ -13,7 +13,7 @@ plot_venn_diagram <- function(agg_table)
 
  # load sql config and get connection
  source("sql_config.R")
- con <- get_sql_conn('/home/patrick/dev/bio/data/test_matching.sqlite')
+ con <- get_sql_conn()
 
 
  # load ppi network from db

diff --git a/data/.gitignore b/data/.gitignore
@@ -8,6 +8,7 @@ tmp/
 # the actual download folder:
 download/
 
+*.sqlite
 
 ################
 # PPIs

diff --git a/src/graph_properties.py b/src/graph_properties.py
@@ -242,8 +242,8 @@ def graph_stats(g, verbose=True, timings={}):
  if verbose:
  print("get connected components")
  t = time.time()
- cc = ppi_networkit.ConnectedComponents()
- cc.run(g)
+ cc = ppi_networkit.ConnectedComponents(g)
+ cc.run()
  props['conn_comp'] = cc.numberOfComponents()
  timings[(ppi_name, "conn_comp")] = time.time() - t
 

diff --git a/src/init_data.py b/src/init_data.py
@@ -50,10 +50,10 @@
 #####################
 
 # load GO terms
-import pappi.go_import
+import pappi.go.import_ass
 
 # import GO associations
-pappi.go_import.import_go_association(GO_ASSOC_FILE, con)
+pappi.go.import_ass.import_go_association(GO_ASSOC_FILE, con)
 
 
 ##############################

diff --git a/src/pappi/data_config.py b/src/pappi/data_config.py
@@ -20,7 +20,7 @@
 MAP_DATA_FOLDER = os.path.join(DATA_FOLDER, 'mapping')
 
 # TODO add these files into the repo data folder !?
-DATABASE = os.path.join(EXT_DATA_FOLDER, 'test_matching.sqlite')
+DATABASE = os.path.join(DATA_FOLDER, 'sql_database.sqlite')
 
 # mapping
 HGNC_FILE = os.path.join(MAP_DATA_FOLDER, 'hgnc_downloads.txt')

diff --git a/src/pappi/go/import.py → src/pappi/go/import_ass.py b/src/pappi/go/import.py → src/pappi/go/import_ass.py
@@ -1,4 +1,4 @@
-from . import sql
+from .. import sql
 
 GO_ASSOC_TABLE = "go_gene_assoc"