(r1) JSlackThesisProposal < Main

Tags: view all tags
---+ Work completed

---++ SJ (2003-2006)

---+++ 2003-2004
SequenceJuxtaposer (*SJ 2004*) is a sequence browsing tool for exploring simiarities between aligned nucleotide sequences.  Input to the browser is in fasta format files, which is a standard used widely by sequence analysis tools.  The original goals of SJ was to provide an accordion drawing navigation tool that combined the tree structures of TreeJuxtaposer (*TJ 2003*) with the SJ view.  The leaves of TJ are sequences, as is each sequence represented horizontally in SJ, and this alignment of tree and sequence view would give a great way of viewing the corresponding relationships between the topology and the sequence data.

However, this project turned rather ambitious after most of the focus of my master's work was based on the underlying navigation and rendering infrastructure (*ADNAV 2006* and *PRISAD 2005*).  The original work of SJ was limited in capability, but also lacked an appropriately well-defined programming interface to communicate with the interface of TJ, which was also limited, perhaps non-existant.

---+++ 2005-2006

The work in *PRISAD 2005* on the rendering improvements for SJ was focused on aggregating data for improving rendering performance.  The caching schemes used similarities between adjacent sequences as an advantage: aligned nucleotides often contain similar or identical nucleotides in a given position.  SJ arranges sequences in rows, so a column of aligned nucleotides is the most efficient direction for the first stage of culling.  A second pass of culling, done on the first rendering pass, creates a hierarchical aggregate column based on the partitioning tree used to perform layout and navigation (the split line hierarchy).  Future rendering, which uses the hierarchy to determine what is large enough to be considered one pixel (or block), uses the cached values to later render columns as required.  The rendering order is also determined with user navigations, so we draw inside user selected regions first, and fill in details if we have enough time to complete the rendering.

---+ Work in progress

---++ SJ2? (fosmid explorer?)

---+++ 2006-

Collaborating with Steven Hallam, who has a lab at UBC, we want to see if we can use accordion drawing, or similar techniques, to explore a different type of dataset.  This is a picture from Hallam et al. (*Reverse Methanogenesis 2004*):

   * fosmid comparison (From Reverse Methanogenesis:..., Steven Hallam et al): <br />
    <img src="%ATTACHURLPATH%/hallamPic.png" alt="hallamPic.png" width="640" />

This figure was drawn "by hand" to show the relationships between 19 different genetic sequences.  The similar parts are connected in each row by pologons, where shading gives directionality: an "X" with dark top and light bottom shows a flip in a particular gene, while a direct correspondance is more of a parallelogram (rectangle where possible).  This is a picture from software developed at the Sanger Center (http://www.sanger.ac.uk/Software/ACT/):

   * blasted pair of sequences, using ACT from the Sanger Center.: <br />
    <img src="%ATTACHURLPATH%/AY714841vsAY714826_blastnOUT.png" alt="blasted pair of sequences, using ACT from the sanger center."  width="640" />

This figure shows different information in the comparison of 2 different sequences.  The similarities are represented in the same way.  The greater details are useful for investigating the sequences from their origins.  However, the scalability of this system is limited.  The length of sequences is restricted by the window size (no zoom function), and it only supports pairwise comparisons.  A more robust interactive system that scales to show the details of the ACT system as well as a high-level "Hallam" view is desirable.
Attachments
Topic attachments
I	Attachment	History	Action	Size	Date	Who	Comment
png	AY714841vsAY714826_blastnOUT.png	r1	manage	57.1 K	2006-11-29 - 17:41	JamesSlack	blasted pair of sequences, using ACT from the sanger center.
png	hallamPic.png	r1	manage	70.4 K	2006-11-29 - 17:30	JamesSlack
Topic revision: r1 - 2006-11-29 - JamesSlack
~~Edit~~
~~Attach~~