trunk/sequence/GTGHitList.java

package sequence;

import java.sql.SQLException;
import java.util.ArrayList;
import java.util.Iterator;

import proteinstructure.PdbCodeNotFoundError;
import proteinstructure.PdbLoadError;
import proteinstructure.TemplateList;

import tools.MySQLConnection;

/**
 * A convenience class for a list of GTGHits
 *
 */
public class GTGHitList implements Iterable<GTGHit> {

        private ArrayList<GTGHit> hits;
        
        private int queryLength;
        
        public GTGHitList() {
                this.hits = new ArrayList<GTGHit>();
        }
        
        public void add(GTGHit hit) {
                hits.add(hit);
        }
        
        public int size() {
                return this.hits.size();
        }
        
        public void printTabular() {
                for (GTGHit hit:this.hits) {
                        hit.print();
                }
        }
        
        /**
         * Returns and iterator over this GTG hits list
         * @return
         */
        public Iterator<GTGHit> iterator() {
                return this.hits.iterator();
        }
        
        /**
         * Reassigns the serials of subjectStart and subjectEnd positions based on the 
         * pdbase full SEQRES sequence, i.e. converts GTG serials to our internal residue
         * serials (cif serials)
         * @param conn
         * @param pdbaseDb
         * @throws SQLException
         * @throws PdbCodeNotFoundError
         * @throws PdbLoadError
         * @see GTGHit.reassignSubjectSerials
         */
        public void reassignSubjectSerials(MySQLConnection conn, String pdbaseDb) throws SQLException, PdbCodeNotFoundError, PdbLoadError {
                for (GTGHit hit: this) {
                        hit.reassignSubjectSerials(conn, pdbaseDb);
                }       
        }
        
        /**
         * Prints a tabular overview of the hits in this list.
         * Currently, if the query length is set, (i.e. > 0) a simple ascii-art
         * overview of the matches is printed for each hit.
         */
        public void print() {
                printSome(this.size());
        }
        
        /**
         * Prints a tabular overview of the first numHits hits in this list.
         * Currently, if the query length is set, (i.e. > 0) a simple ascii-art
         * overview of the matches is printed for each hit.
         * @param numHits
         */
        public void printSome(int numHits) {
                
                int outputLength = 80;          // length of graphical output in screen columns
                
                if(queryLength > 0) {
                        double scaleFactor = 1.0 * outputLength / queryLength;
                        GTGHit.printHeaderWithOverview(queryLength, scaleFactor);
                        for (int i = 0; i < Math.min(hits.size(), numHits); i++) {
                                GTGHit hit = hits.get(i);
                                hit.printWithOverview(scaleFactor);
                        }                       
                } else {
                        // print without graphical overview
                        for (int i = 0; i < Math.min(hits.size(), numHits); i++) {
                                GTGHit hit = hits.get(i);
                                hit.print();
                        }                       
                }
        }
        
        /**
         * Set the query length needed by the print() and printSome() methods.
         * @param l
         */
        public void setQueryLength(int l) {
                this.queryLength = l;
        }
        
        /**
         * Applies an total score cutoff trimming out of this list all hits with totalScore 
         * lower than given cutoff
         * @param totalScoreCutoff
         */
        public void applyCutoff(int totalScoreCutoff) {
                Iterator<GTGHit> it = this.iterator();
                while (it.hasNext()) {
                        if (it.next().totalScore<=totalScoreCutoff) {
                                it.remove();
                        }
                }
        }
        
        /**
         * Filters out from this GTGHitList PDB hits released after the given date
         * (in format yyyymmdd, e.g. 20060425)  
         * @param conn
         * @param pdbaseDb
         * @param date date in format yyyymmdd
         * @throws SQLException
         */
        public void filterByMaxReleaseDate(MySQLConnection conn, String pdbaseDb, String date) throws SQLException {
                String[] filteredIds = TemplateList.filterIdsByMaxReleaseDate(conn, pdbaseDb, this.getTemplateIds(), date);
                for (String filteredId:filteredIds) {
                        this.hits.remove(this.getHit(filteredId));
                }
        }
        
        /**
         * Filters out hits that rank below given rank.
         * @param rank
         */
        public void filterByMaxRank(int rank) {
                Iterator<GTGHit> it = this.iterator();
                int i=1;
                while (it.hasNext()) {
                        it.next();
                        if (i>rank) it.remove();
                        i++;
                }
        }

        /**
         * Returns the hit with the best totalScore or null if this hit list is empty.
         * @return
         */
        public GTGHit getBestHit() {
                if(this.size() == 0) return null;
                GTGHit bestHit = this.hits.get(0);
                for(GTGHit hit:hits) {
                        if(hit.totalScore > bestHit.totalScore) {
                                bestHit = hit;
                        }
                }
                return bestHit;
        }
        
        /**
         * Gets a GTGHit given its subjectId or null if GTGHit not present in this list 
         * @param templateId
         * @return
         */
        public GTGHit getHit(String templateId) {
                for (GTGHit hit:this) {
                        if (hit.getTemplateId().equals(templateId)) return hit;
                }
                return null;
        }
        
        /**
         * Returns a list of template ids (subject ids) as concatenated 
         * pdbCodes+pdbChainCodes e.g. 1abcA 
         * @return
         */
        public String[] getTemplateIds() {
                String[] ids = new String[this.size()];
                for (int i=0;i<this.size();i++) {
                        ids[i] = this.hits.get(i).getTemplateId();
                }
                return ids;
        }
        
}
Revision:	622
Committed:	Fri May 9 12:56:27 2008 UTC (16 years, 5 months ago) by duarte
File size:	4679 byte(s)
Log Message:	Added code to be able to filter out template ids (pdb codes) by maximum release date. Some improvements in Hit classes.
Line	File contents
1	package sequence;
2
3	import java.sql.SQLException;
4	import java.util.ArrayList;
5	import java.util.Iterator;
6
7	import proteinstructure.PdbCodeNotFoundError;
8	import proteinstructure.PdbLoadError;
9	import proteinstructure.TemplateList;
10
11	import tools.MySQLConnection;
12
13	/**
14	* A convenience class for a list of GTGHits
15	*
16	*/
17	public class GTGHitList implements Iterable<GTGHit> {
18
19	private ArrayList<GTGHit> hits;
20
21	private int queryLength;
22
23	public GTGHitList() {
24	this.hits = new ArrayList<GTGHit>();
25	}
26
27	public void add(GTGHit hit) {
28	hits.add(hit);
29	}
30
31	public int size() {
32	return this.hits.size();
33	}
34
35	public void printTabular() {
36	for (GTGHit hit:this.hits) {
37	hit.print();
38	}
39	}
40
41	/**
42	* Returns and iterator over this GTG hits list
43	* @return
44	*/
45	public Iterator<GTGHit> iterator() {
46	return this.hits.iterator();
47	}
48
49	/**
50	* Reassigns the serials of subjectStart and subjectEnd positions based on the
51	* pdbase full SEQRES sequence, i.e. converts GTG serials to our internal residue
52	* serials (cif serials)
53	* @param conn
54	* @param pdbaseDb
55	* @throws SQLException
56	* @throws PdbCodeNotFoundError
57	* @throws PdbLoadError
58	* @see GTGHit.reassignSubjectSerials
59	*/
60	public void reassignSubjectSerials(MySQLConnection conn, String pdbaseDb) throws SQLException, PdbCodeNotFoundError, PdbLoadError {
61	for (GTGHit hit: this) {
62	hit.reassignSubjectSerials(conn, pdbaseDb);
63	}
64	}
65
66	/**
67	* Prints a tabular overview of the hits in this list.
68	* Currently, if the query length is set, (i.e. > 0) a simple ascii-art
69	* overview of the matches is printed for each hit.
70	*/
71	public void print() {
72	printSome(this.size());
73	}
74
75	/**
76	* Prints a tabular overview of the first numHits hits in this list.
77	* Currently, if the query length is set, (i.e. > 0) a simple ascii-art
78	* overview of the matches is printed for each hit.
79	* @param numHits
80	*/
81	public void printSome(int numHits) {
82
83	int outputLength = 80; // length of graphical output in screen columns
84
85	if(queryLength > 0) {
86	double scaleFactor = 1.0 * outputLength / queryLength;
87	GTGHit.printHeaderWithOverview(queryLength, scaleFactor);
88	for (int i = 0; i < Math.min(hits.size(), numHits); i++) {
89	GTGHit hit = hits.get(i);
90	hit.printWithOverview(scaleFactor);
91	}
92	} else {
93	// print without graphical overview
94	for (int i = 0; i < Math.min(hits.size(), numHits); i++) {
95	GTGHit hit = hits.get(i);
96	hit.print();
97	}
98	}
99	}
100
101	/**
102	* Set the query length needed by the print() and printSome() methods.
103	* @param l
104	*/
105	public void setQueryLength(int l) {
106	this.queryLength = l;
107	}
108
109	/**
110	* Applies an total score cutoff trimming out of this list all hits with totalScore
111	* lower than given cutoff
112	* @param totalScoreCutoff
113	*/
114	public void applyCutoff(int totalScoreCutoff) {
115	Iterator<GTGHit> it = this.iterator();
116	while (it.hasNext()) {
117	if (it.next().totalScore<=totalScoreCutoff) {
118	it.remove();
119	}
120	}
121	}
122
123	/**
124	* Filters out from this GTGHitList PDB hits released after the given date
125	* (in format yyyymmdd, e.g. 20060425)
126	* @param conn
127	* @param pdbaseDb
128	* @param date date in format yyyymmdd
129	* @throws SQLException
130	*/
131	public void filterByMaxReleaseDate(MySQLConnection conn, String pdbaseDb, String date) throws SQLException {
132	String[] filteredIds = TemplateList.filterIdsByMaxReleaseDate(conn, pdbaseDb, this.getTemplateIds(), date);
133	for (String filteredId:filteredIds) {
134	this.hits.remove(this.getHit(filteredId));
135	}
136	}
137
138	/**
139	* Filters out hits that rank below given rank.
140	* @param rank
141	*/
142	public void filterByMaxRank(int rank) {
143	Iterator<GTGHit> it = this.iterator();
144	int i=1;
145	while (it.hasNext()) {
146	it.next();
147	if (i>rank) it.remove();
148	i++;
149	}
150	}
151
152	/**
153	* Returns the hit with the best totalScore or null if this hit list is empty.
154	* @return
155	*/
156	public GTGHit getBestHit() {
157	if(this.size() == 0) return null;
158	GTGHit bestHit = this.hits.get(0);
159	for(GTGHit hit:hits) {
160	if(hit.totalScore > bestHit.totalScore) {
161	bestHit = hit;
162	}
163	}
164	return bestHit;
165	}
166
167	/**
168	* Gets a GTGHit given its subjectId or null if GTGHit not present in this list
169	* @param templateId
170	* @return
171	*/
172	public GTGHit getHit(String templateId) {
173	for (GTGHit hit:this) {
174	if (hit.getTemplateId().equals(templateId)) return hit;
175	}
176	return null;
177	}
178
179	/**
180	* Returns a list of template ids (subject ids) as concatenated
181	* pdbCodes+pdbChainCodes e.g. 1abcA
182	* @return
183	*/
184	public String[] getTemplateIds() {
185	String[] ids = new String[this.size()];
186	for (int i=0;i<this.size();i++) {
187	ids[i] = this.hits.get(i).getTemplateId();
188	}
189	return ids;
190	}
191
192	}