From f4506ad5ac1f43a8d00951a91f07d9f51b75b8a4 Mon Sep 17 00:00:00 2001
From: Natasha Sherstneva <n.shertneva@gmail.com>
Date: Thu, 7 Nov 2013 13:32:02 +0000
Subject: [PATCH] fix data report, protein report

Conflicts:
	.classpath
	datadb/compbio/cassandra/CassandraNativeConnector.java
---
 .../cassandra/CassandraNativeConnector.java        |  105 +++++++++++++++++++-
 1 file changed, 102 insertions(+), 3 deletions(-)

diff --git a/datadb/compbio/cassandra/CassandraNativeConnector.java b/datadb/compbio/cassandra/CassandraNativeConnector.java
index 766c376..8e86624 100644
--- a/datadb/compbio/cassandra/CassandraNativeConnector.java
+++ b/datadb/compbio/cassandra/CassandraNativeConnector.java
@@ -339,6 +339,105 @@ public class CassandraNativeConnector {
 		System.out.println (c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
 		return res;
 	}
+	
+	/*
+	 * getting data from the db ProteinData
+	 */
+	public Integer ReadDateTable(long queryDate) {
+		final long startTime = System.currentTimeMillis();
+		String com = "SELECT jobtime, JobID FROM ProteinKeyspace.ProteinData WHERE jobtime = " + queryDate + ";";
+		System.out.println("Command: " + com);
+		ResultSet results = session.execute(com);
+		if (results.isExhausted())
+			return null;
+		final long queryTime = System.currentTimeMillis();
+		List<Row> rows = results.all();
+		System.out.println ("Query time is " + (queryTime - startTime) + " msec");        
+		return rows.size();
+	}
+
+	/*
+	 * getting whole protein sequence from the db ProteinRow
+	 */
+	public List<StructureProteinPrediction> ReadWholeSequence(String queryProtein) {
+		final long startTime = System.currentTimeMillis();
+		String com = "SELECT JobID, Predictions FROM ProteinKeyspace.ProteinRow WHERE Protein = '" + queryProtein + "';";
+		System.out.println("Command: " + com);
+		ResultSet results = session.execute(com);
+		if (results.isExhausted())
+			return null;
+		final long queryTime = System.currentTimeMillis();
+		List<Row> rows = results.all();
+		System.out.println ("Query time is " + (queryTime - startTime) + " msec");   
+		System.out.println (" rows analysed,  " + rows.size());
+		List<StructureProteinPrediction> res = new ArrayList<StructureProteinPrediction>();
+		int c = 0;
+		for (Row r : rows) {
+			StructureProteinPrediction structure = new StructureProteinPrediction(queryProtein, r.getString("JobID"), r.getMap("Predictions", String.class, String.class));		
+			res.add(structure);
+			++c;
+		}
+		final long endTime = System.currentTimeMillis();
+		System.out.println (c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
+		return res;
+	}
+	
+	/*
+	 * getting part of protein sequence from the db ProteinRow
+	 */
+	public List<StructureProteinPrediction>  ReadPartOfSequence(String queryProtein) {
+		final long startTime = System.currentTimeMillis();
+		String com = "SELECT * FROM ProteinKeyspace.ProteinRow;";
+		System.out.println("Command: " + com);
+		ResultSet results = session.execute(com);
+		if (results.isExhausted())
+			return null;
+		final long queryTime = System.currentTimeMillis();
+		List<Row> rows = results.all();
+		System.out.println ("Query time is " + (queryTime - startTime) + " msec");   
+		System.out.println (" rows analysed,  " + rows.size());
+		List<StructureProteinPrediction>  res = new ArrayList<StructureProteinPrediction>();
+		int c = 0;
+		for (Row r : rows) {
+			String prot = r.getString("Protein");
+			if (prot.matches("(.*)" + queryProtein + "(.*)")) {
+				StructureProteinPrediction structure = new StructureProteinPrediction(prot, r.getString("JobID"), r.getMap("Predictions", String.class, String.class));		
+				res.add(structure);
+				++c;
+			}
+		}
+		final long endTime = System.currentTimeMillis();
+		System.out.println (c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
+		return res;
+	}
+	
+	/*
+	 * getting protein sequences by counter
+	 */
+	public Map<String, Integer>  ReadProteinDataByCounter() {
+		final long startTime = System.currentTimeMillis();
+		String com = "SELECT Protein FROM ProteinKeyspace.ProteinRow;";
+		System.out.println("Command: " + com);
+		ResultSet results = session.execute(com);
+		if (results.isExhausted())
+			return null;
+		final long queryTime = System.currentTimeMillis();
+		List<Row> rows = results.all();
+		System.out.println ("Query time is " + (queryTime - startTime) + " msec");   
+		System.out.println (" rows analysed,  " + rows.size());
+		Map<String, Integer> res = new HashMap<String, Integer>();
+		int c = 0;
+		for (Row r : rows) {
+			String protein = r.getString("Protein");
+			if (res.containsKey(protein)) 
+				res.put(protein, res.get(protein) + 1);
+			else
+				res.put(protein, 1);
+		}
+		final long endTime = System.currentTimeMillis();
+		System.out.println (c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
+		return res;
+	}
 
 	/*
 	 * getting protein sequences by counter
@@ -365,7 +464,7 @@ public class CassandraNativeConnector {
 		return res;
 	}
 
-	/*
+
 	 * getting earlest date of jobs from the db
 	 */
 	public long getEarliestDateInDB() {
@@ -374,7 +473,7 @@ public class CassandraNativeConnector {
 		System.out.println("Command: " + com);
 		ResultSet results = session.execute(com);
 		final long queryTime = System.currentTimeMillis();
-		System.out.println("Query time is " + (queryTime - startTime) + " msec");
+		System.out.println ("Query time is  " + (queryTime - startTime) + " msec");
 
 		Calendar cal = Calendar.getInstance();
 		long res = cal.getTimeInMillis();
@@ -388,7 +487,7 @@ public class CassandraNativeConnector {
 			++c;
 		}
 		final long endTime = System.currentTimeMillis();
-		System.out.println(c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
+		System.out.println (c + " rows analysed, execution time is " + (endTime - startTime) + " msec");
 		return res;
 	}
 
-- 
1.7.10.2