Merge branch 'develop' of https://source.jalview.org/git/jalview into develop

[jalview.git] / src / jalview / analysis / Profile.java

diff --git a/src/jalview/analysis/Profile.java b/src/jalview/analysis/Profile.java
new file mode 100644 (file)
index 0000000..d94d031
--- /dev/null
+++ b/src/jalview/analysis/Profile.java
@@ -0,0 +1,157 @@
+package jalview.analysis;
+
+
+/**
+ * A data bean to hold the result of computing a profile for a column of an
+ * alignment
+ * 
+ * @author gmcarstairs
+ *
+ */
+public class Profile
+{
+  /*
+   * counts of keys (chars)
+   */
+  private ResidueCount counts;
+
+  /*
+   * the number of sequences in the profile
+   */
+  private int height;
+
+  /*
+   * the number of non-gapped sequences in the profile
+   */
+  private int gapped;
+
+  /*
+   * the highest count for any residue in the profile
+   */
+  private int maxCount;
+
+  /*
+   * the residue (e.g. K) or residues (e.g. KQW) with the
+   * highest count in the profile
+   */
+  private String modalResidue;
+
+  /**
+   * Constructor which allows derived data to be stored without having to store
+   * the full profile
+   * 
+   * @param seqCount
+   *          the number of sequences in the profile
+   * @param gaps
+   *          the number of gapped sequences
+   * @param max
+   *          the highest count for any residue
+   * @param modalres
+   *          the residue (or concatenated residues) with the highest count
+   */
+  public Profile(int seqCount, int gaps, int max, String modalRes)
+  {
+    this.height = seqCount;
+    this.gapped = gaps;
+    this.maxCount = max;
+    this.modalResidue = modalRes;
+  }
+
+  /**
+   * Set the full profile of counts
+   * 
+   * @param residueCounts
+   */
+  public void setCounts(ResidueCount residueCounts)
+  {
+    this.counts = residueCounts;
+  }
+
+  /**
+   * Returns the percentage identity of the profile, i.e. the highest proportion
+   * of conserved (equal) symbols. The percentage is as a fraction of all
+   * sequences, or only ungapped sequences if flag ignoreGaps is set true.
+   * 
+   * @param ignoreGaps
+   * @return
+   */
+  public float getPercentageIdentity(boolean ignoreGaps)
+  {
+    if (height == 0)
+    {
+      return 0f;
+    }
+    float pid = 0f;
+    if (ignoreGaps && gapped < height)
+    {
+      pid = (maxCount * 100f) / (height - gapped);
+    }
+    else
+    {
+      pid = (maxCount * 100f) / height;
+    }
+    return pid;
+  }
+
+  /**
+   * Returns the full symbol counts for this profile
+   * 
+   * @return
+   */
+  public ResidueCount getCounts()
+  {
+    return counts;
+  }
+
+  /**
+   * Returns the number of sequences in the profile
+   * 
+   * @return
+   */
+  public int getHeight()
+  {
+    return height;
+  }
+
+  /**
+   * Returns the number of sequences in the profile which had a gap character
+   * (or were too short to be included in this column's profile)
+   * 
+   * @return
+   */
+  public int getGapped()
+  {
+    return gapped;
+  }
+
+  /**
+   * Returns the highest count for any symbol(s) in the profile
+   * 
+   * @return
+   */
+  public int getMaxCount()
+  {
+    return maxCount;
+  }
+
+  /**
+   * Returns the symbol (or concatenated symbols) which have the highest count
+   * in the profile, or an empty string if there were no symbols counted
+   * 
+   * @return
+   */
+  public String getModalResidue()
+  {
+    return modalResidue;
+  }
+
+  /**
+   * Answers the number of non-gapped sequences in the profile
+   * 
+   * @return
+   */
+  public int getNonGapped()
+  {
+    return height - gapped;
+  }
+}