import jalview.datamodel.AlignmentI;
import jalview.datamodel.SequenceGroup;
import jalview.datamodel.SequenceI;
+import jalview.ws.params.OptionI;
+import jalview.ws.params.simple.IntegerParameter;
+import jalview.ws.params.simple.Option;
import jalview.ws.rest.AlignmentProcessor;
import jalview.ws.rest.InputType;
import jalview.ws.rest.NoValidInputDataException;
import jalview.ws.rest.RestJob;
+import jalview.ws.rest.RestServiceDescription;
import jalview.ws.rest.InputType.molType;
import java.io.UnsupportedEncodingException;
import java.util.ArrayList;
+import java.util.Arrays;
+import java.util.List;
import java.util.Vector;
import org.apache.http.entity.mime.content.ContentBody;
import org.apache.http.entity.mime.content.StringBody;
/**
- * Represents the partitions defined on the alignment as indices
- * e.g. for a partition (A,B,C),(D,E),(F)
- * The indices would be 3,2,1. Note, the alignment must be ordered so groups are contiguous before this input type can be used.
+ * Represents the partitions defined on the alignment as indices e.g. for a
+ * partition (A,B,C),(D,E),(F) The indices would be 3,2,1. Note, the alignment
+ * must be ordered so groups are contiguous before this input type can be used.
+ *
* @author JimP
- *
+ *
*/
-public class SeqGroupIndexVector extends InputType implements AlignmentProcessor{
+public class SeqGroupIndexVector extends InputType implements
+ AlignmentProcessor
+{
public SeqGroupIndexVector()
{
- super(new Class[] { AlignmentI.class} );
+ super(new Class[]
+ { AlignmentI.class });
}
+
/**
* separator for list of sequence Indices - default is ','
*/
- public String sep=",";
+ public String sep = ",";
+
+ /**
+ * min size of each partition
+ */
+ public int minsize = 1;
+
molType type;
+
/**
* prepare the context alignment for this input
- * @param al - alignment to be processed
+ *
+ * @param al
+ * - alignment to be processed
* @return al or a new alignment with appropriate attributes/order for input
*/
public AlignmentI prepareAlignment(AlignmentI al)
jalview.analysis.AlignmentSorter.sortByGroup(al);
return al;
}
+
@Override
- public ContentBody formatForInput(RestJob rj) throws UnsupportedEncodingException, NoValidInputDataException
+ public ContentBody formatForInput(RestJob rj)
+ throws UnsupportedEncodingException, NoValidInputDataException
{
StringBuffer idvector = new StringBuffer();
- boolean list=false;
+ boolean list = false;
AlignmentI al = rj.getAlignmentForInput(token, type);
- // assume that alignment is properly ordered so groups form consecutive blocks
+ // assume that alignment is properly ordered so groups form consecutive
+ // blocks
ArrayList<int[]> gl = new ArrayList<int[]>();
- for (SequenceGroup sg : (Vector<SequenceGroup>)al.getGroups())
+ int p = 0;
+ for (SequenceGroup sg : (Vector<SequenceGroup>) al.getGroups())
{
- // TODO: refactor to sequenceGroup for efficiency - getAlignmentRowInterval(AlignmentI al)
- int[] se=null;
- for (SequenceI sq: sg.getSequencesInOrder(al))
+ if (sg.getSize() < minsize)
{
- int p = al.findIndex(sq);
- if (se==null)
+ throw new NoValidInputDataException("Group contains less than "
+ + minsize + " sequences.");
+ }
+ // TODO: refactor to sequenceGroup for efficiency -
+ // getAlignmentRowInterval(AlignmentI al)
+ int[] se = null;
+ for (SequenceI sq : sg.getSequencesInOrder(al))
+ {
+ p = al.findIndex(sq);
+ if (se == null)
{
- se=new int[] { p, p};
+ se = new int[]
+ { p, p };
}
- else {
- if (p<se[0])
- se[0]=p;
- if (p>se[1])
- se[1]=p;
+ else
+ {
+ if (p < se[0])
+ se[0] = p;
+ if (p > se[1])
+ se[1] = p;
}
}
- if (se!=null)
+ if (se != null)
{
gl.add(se);
}
}
+ // are there any more sequences ungrouped that should be added as a single
+ // remaining group ? - these might be at the start or the end
+ if (gl.size() > 0)
+ {
+ int[] tail = gl.get(0);
+ if (tail[0] > 0)
+ {
+ if (1 + tail[0] > minsize)
+ {
+ gl.add(0, new int[]
+ { 0, tail[0] - 1 });
+ }
+ else
+ {
+ // lets be intelligent here - if the remaining sequences aren't enough
+ // to make a final group, then don't make one.
+ // throw new
+ // NoValidInputDataException("Group from remaining ungrouped sequences in input contains less than "+minsize+" sequences.");
+ }
+ }
+ else
+ {
+ tail = gl.get(gl.size() - 1);
+ if (1 + tail[1] < al.getHeight())
+ {
+ if (al.getHeight() - (1 + tail[1]) > minsize)
+ {
+ gl.add(new int[]
+ { tail[1] + 1, al.getHeight() - 1 });
+ }
+ else
+ {
+ // lets be intelligent here - if the remaining sequences aren't
+ // enough to make a final group, then don't make one.
+ // throw new
+ // NoValidInputDataException("Group from remaining ungrouped sequences in input contains less than "+minsize+" sequences.");
+ }
+ }
+ }
+ }
+ else
+ {
+ gl.add(new int[]
+ { 0, al.getHeight() - 1 });
+ }
+ if (min >= 0 && gl.size() < min)
+ {
+ throw new NoValidInputDataException(
+ "Not enough sequence groups for input. Need at least " + min
+ + " groups (including ungrouped regions).");
+ }
+ if (max > 0 && gl.size() > max)
+ {
+ throw new NoValidInputDataException(
+ "Too many sequence groups for input. Need at most " + max
+ + " groups (including ungrouped regions).");
+ }
int[][] vals = gl.toArray(new int[gl.size()][]);
int[] srt = new int[gl.size()];
- for (int i=0;i<vals.length;i++)
- srt[i]=vals[i][0];
+ for (int i = 0; i < vals.length; i++)
+ srt[i] = vals[i][0];
jalview.util.QuickSort.sort(srt, vals);
- list=false;
- int last=vals[0][0];
- for (int[] range:vals)
+ list = false;
+ int last = vals[0][0] - 1;
+ for (int[] range : vals)
{
- if (range[1]>last) {
+ if (range[1] > last)
+ {
if (list)
+ {
+ idvector.append(sep);
+ }
+ idvector.append(range[1] - last);
+ last = range[1];
+ list = true;
+ }
+ }
+ return new StringBody(idvector.toString());
+ }
+
+ /**
+ * set minimum number of sequences allowed in a partition. Default is 1
+ * sequence.
+ *
+ * @param i
+ * (number greater than 1)
+ */
+ public void setMinsize(int i)
+ {
+ if (minsize >= 1)
+ {
+ minsize = i;
+ }
+ else
+ {
+ minsize = 1;
+ }
+ }
+
+ @Override
+ public List<String> getURLEncodedParameter()
+ {
+ ArrayList<String> prms = new ArrayList<String>();
+ super.addBaseParams(prms);
+ prms.add("minsize='" + minsize + "'");
+ prms.add("sep='" + sep + "'");
+ if (type != null)
+ {
+ prms.add("type='" + type + "'");
+ }
+ return prms;
+ }
+
+ @Override
+ public String getURLtokenPrefix()
+ {
+ return "PARTITION";
+ }
+
+ @Override
+ public boolean configureProperty(String tok, String val,
+ StringBuffer warnings)
+ {
+
+ if (tok.startsWith("sep"))
+ {
+ sep = val;
+ return true;
+ }
+ if (tok.startsWith("minsize"))
+ {
+ try
+ {
+ minsize = Integer.valueOf(val);
+ if (minsize >= 0)
+ return true;
+ } catch (Exception x)
{
- idvector.append(sep);
+
}
- idvector.append(range[1]-last+1);
- last=range[1];
- list=true;
+ warnings.append("Invalid minsize value '" + val
+ + "'. Must be a positive integer.\n");
+ }
+ if (tok.startsWith("type"))
+ {
+ try
+ {
+ type = molType.valueOf(val);
+ return true;
+ } catch (Exception x)
+ {
+ warnings.append("Invalid molecule type '" + val
+ + "'. Must be one of (");
+ for (molType v : molType.values())
+ {
+ warnings.append(" " + v);
+ }
+ warnings.append(")\n");
}
}
- return new StringBody(idvector.toString());
+ return false;
}
-
-}
\ No newline at end of file
+
+ @Override
+ public List<OptionI> getOptions()
+ {
+ List<OptionI> lst = getBaseOptions();
+ lst.add(new Option("sep",
+ "Separator character between elements of vector", true, ",",
+ sep, Arrays.asList(new String[]
+ { " ", ",", ";", "\t", "|" }), null));
+ lst.add(new IntegerParameter("minsize", "Minimum size of partition allowed by service", true, 1, minsize, 1,0));
+ lst.add(createMolTypeOption("type", "Sequence type", false, type, molType.MIX));
+ return lst;
+ }
+
+}
git://source.jalview.org / jalview.git / blobdiff