:alt: New Mussa Window
:align: center
- A new Mussa window on the right, in which I have loaded a second
- experiment.
+ A new Mussa window on the right, in which a second analysis has
+ been loaded.
Now you can create or load an existing analysis, in this new window,
as described in the `Create/Load Analysis`_ section.
To run a sub-analysis **highlight** a section of sequence and *right
click* on it and select **Add to subanalysis**. To the same for the
sequences shown in orange in the screenshot below. Note that you **are
-NOT limited** to selecting more than one subsequence from the same
+NOT limited** to selecting only one subsequence from the same
sequence.
.. image:: images/subanalysis_select_seqs.png
Saving to an Image
---------------------------------
- * Updated to build 419.
-
To save your current mussa view to an image, select **File > Save to
image...** as shown below.
====== ================= ===================================
+
+Understanding Mussa
+===================
+
+
+Performance
+-----------
+
+Algorithm Behavior
+~~~~~~~~~~~~~~~~~~
+
+FIXME: Include seqcomp algorithm info.
+
+FIXME: Include transitivity info.
+
+Repeats
+~~~~~~~
+
+The algorithm Mussa uses to find conserved sequences is sensative to
+repeated DNA segments, which are naturally apart of many genomes. The
+problem with repeats, is that one repeat from one sequence can show up
+many times in another sequence. Every connection Mussa makes takes up
+memory, and it also takes time to store and process the results.
+
+The formula for the number of connections, C, that will be made for R
+instances of a single repeat (meaning R copies of one repeat in each
+sequence) and S sequences is:
+
+C = (R^2)[S(S-1)/2]
+
+Table of example situations:
+
+===== ===== =====
+ C R S
+===== ===== =====
+ 16 4 2
+ 48 4 3
+ 96 4 4
+ 160 4 5
+ 240 4 6
+ 336 4 7
+ 448 4 8
+ 24 2 4
+ 54 3 4
+ 96 4 4
+ 150 5 4
+ 216 6 4
+ 294 7 4
+ 384 8 4
+ 2500 50 2
+ 7500 50 3
+15000 50 4
+10000 100 2
+30000 100 3
+60000 100 4
+===== ===== =====
+
+After the connections, C, are found, they are passed on to the
+transitivity filter, which is a C^2 algorithm (FIXME: confirm
+algorithm is C^2). This means with 50 repeats in 2 sequences giving
+you a C of 2500, ends up with a C^2 of 6,250,000.
+
+**Conclusion: repeats cause the processing time of Mussa to skyrocket.**
+
+One, way to deal with a situation where you have lots of repeats in
+your sequences is to use shorter sequences lengths and/or repeat mask
+at least one of your sequences.
+
+Details
+-------
+
+Case: Conservation track suddenly stops
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+
.. Define links below
------------------