X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=datadb%2Fcompbio%2Fcassandra%2Freaders%2FSequenceReader.java;fp=datadb%2Fcompbio%2Fcassandra%2Freaders%2FSequenceReader.java;h=7e77126c6c2f7c55cd68dcf8e60db0eddbbfd57f;hb=9bb6ee99ca7f738fac1087190b5481b8fe6e8d9f;hp=0000000000000000000000000000000000000000;hpb=2e3f6b76be585306f1003d849831840c0adb3360;p=proteocache.git

diff --git a/datadb/compbio/cassandra/readers/SequenceReader.java b/datadb/compbio/cassandra/readers/SequenceReader.java
new file mode 100644
index 0000000..7e77126
--- /dev/null
+++ b/datadb/compbio/cassandra/readers/SequenceReader.java
@@ -0,0 +1,98 @@
+package compbio.cassandra.readers;
+
+import java.util.ArrayList;
+import java.util.List;
+
+import com.datastax.driver.core.ResultSet;
+import com.datastax.driver.core.Row;
+
+import compbio.beans.ProteinBean;
+
+/**
+ * Reader class for making requests on protein sequences to cassandra.
+ * 
+ * @author Alexander Sherstnev
+ * @author Natasha Sherstneva
+ * 
+ * @since 0.5
+ * @version 1.0 
+ * @since December 2013
+ */
+public class SequenceReader extends CassandraReader {
+
+	public SequenceReader() {
+		super();
+	}
+
+	/**
+	 * query: protein sequence
+	 * 
+	 * @param sequence
+	 *            protein sequence or partial of protein sequence
+	 * @param searchtype
+	 *            "whole" or "partial" of protein sequence
+	 * 
+	 * @return List<ProteinBean> to the controller SequenceController
+	 * 
+	 **/
+	public List<ProteinBean> readProteins(String sequence, String searchtype) {
+		List<ProteinBean> query = new ArrayList<ProteinBean>();
+		if (searchtype.equals("whole")) {
+			ResultSet results = CassandraQuery("SELECT JobID, Predictions FROM ProteinRow WHERE Protein = '" + sequence + "';");
+			if (results.isExhausted())
+				return null;
+			List<Row> rows = results.all();
+			ProteinBean structure = new ProteinBean(sequence, rows.get(0).getMap("Predictions", String.class, String.class));
+			for (Row r : rows) {
+				structure.setJobid(r.getString("JobID"));
+			}
+			query.add(structure);
+		} else {
+			ResultSet results = CassandraQuery("SELECT * FROM ProteinRow;");
+			if (results.isExhausted())
+				return null;
+			List<Row> rows = results.all();
+			for (Row r : rows) {
+				String protein = r.getString("Protein");
+				if (protein.matches("(.*)" + sequence + "(.*)")) {
+					ProteinBean foundsequence = new ProteinBean(protein, r.getMap("Predictions", String.class, String.class));
+					foundsequence.setJobid(r.getString("JobID"));
+					query.add(foundsequence);
+				}
+			}
+		}
+		if (searchtype.equals("partial")) {
+			for (ProteinBean entry : query) {
+				entry.setSubProt(CreateSubprotein(entry.getSequence(), sequence));
+			}
+		}
+		return query;
+	}
+
+	/**
+	 * create a list of parts of protein sequence for highlighting current value
+	 * in report;
+	 * 
+	 * @param protein
+	 *            protein sequence
+	 * @param subprot
+	 *            partial of protein sequence
+	 * 
+	 * @return List<String>
+	 * 
+	 **/
+	private static List<String> CreateSubprotein(String protein, String subprot) {
+		List<String> sub = new ArrayList<String>();
+		String subStr = protein;
+		while (subStr.length() > 0 && subStr.contains(subprot)) {
+			String first = subStr.substring(0, subStr.indexOf(subprot));
+			if (first.length() > 0)
+				sub.add(first);
+			sub.add(subprot);
+			subStr = subStr.substring(subStr.indexOf(subprot) + subprot.length(), subStr.length());
+		}
+		if (subStr.length() > 0)
+			sub.add(subStr);
+		return sub;
+	}
+}