About
"The Daily Vic" is an all-you-can-read offering from Victor Parmar, a graduate student at McGill university, Montréal.
Victor is a friendly computer science major who loves traveling, cooking and learning new languages.
If you'd like to contact vic, you can do so here : victorparmar (at) gmail (dot) com. Enjoy your stay!
Designs
I used to have a lot more, but I seem to have lost a lot of my sketches.
- MIT homepage design (first published 25th April 2005!)
- Campus Police!
- Women in Computer Science?
- JSP Web programming course project requirements and home page.
Learn french in montreal
If you are a canadian citizen or a permanent resident, you can avail of the french courses offered by the quebec government for next to nothing: list of CSDM centres offering french as a second language. If you call any of these centres, press '0' to reach a person and in case of a mailbox, leave a message and they will get back to you. For more details regarding Centre St. Louis, check out Jamie King's page.
If you are not a citizen or a permanent resident, the next best (cheapest) places to learn french would be:
- YMCA Language centre located downtown: the standard french courses are 6 hrs/week (usually tuesdays and thursdays from 6-9pm) and cost $315 for 42 hours total.
- Dawson College. I do not know the details, but the courses are good value for money from what I have heard.
If you know of any other good places for learning french in montreal, please let me know and I will update this article with the information.
d3Data
In the information age, data plays a very critical role in shaping how we live our lives, from research to the products and choices that we are presented with. Visualizing data is therefore something that everyone can benefit from, check out some cool stuff here (note that you will need a real browser for this).
Montreal HackReduce
The annual Hadoop MapReduce hackathon was a an amazing time and this bixi heat map is what came out of it.
Thesis: Predicting transcription factor binding sites using phylogenetic footprinting and a probabilistic framework for evolutionary turnover
Identifying genomic locations of transcription-factor binding sites (TFBS), particularly in higher eukaryotic genomes, has been an enormous challenge. Computational methods involving identification of sequence conservation between related genomes have been the most successful since sites found in such highly conserved regions are more likely to be functional, i.e. are bound and regulate protein production. In this thesis, we present such a probabilistic algorithm for predicting TFBSs which also takes evolutionary turnovers into account. Our algorithm is validated via simulations and the results of its application on ChIP-chip data are presented.
In plain english
There are certain proteins, called Transcription Factors which attach on the genome to start the production of other proteins (also known as Transcription). The sequences that these proteins bind are are called Transcription Factor Binding Sites (TFBSs) and they are generally very short, around (6-20) base-pairs on the average. The human genome being 3 billion base-pairs in length, any random 10 base-pair sequence will give 2 million sites given that there are 4 possible base-pairs: Adenine (A), Cytosine (C), Guanine (G) and Thymine (T). So the question is how can we filter out the sites that are actually taking part in the Transcription process? A successful approach, called phylogenetic footprinting, involved comparing the human genome to other mammalian genomes and determining conserved regions. Sites within these regions were probably functional since they are crucial to the organism and hence would not have been lost during evolution. The accuracy of this comparative approach was improved by adding reconstructed intermediate ancestral sequences as well.
Phylogenetic footprinting works well when the sequence composition of a TFBS is fixed, for e.g. ACTTTGTA. However, in reality the Transcription Factors bind to a family of sequences, for e.g. ACTTTGNN, where N means that any of A, C, G or T could be accepted. This not only complicates the site matching, but it also leads to a recently discovered phenomenon called binding site turnover. Turnover can be explained with a simple example: suppose there are two genomes, human and its ancestor aligned with a site for Transcription Factor X at position 500 on each. However, sometime during evolution from the ancestor to the mouse genome one base-pair mutated thereby rendering the site at position 500 non-functional. Since this site is really important to survival, a mutation in the sequence around the original site lead to the creation of a functional site in the mouse genome at position 510. Thus, when the human and mouse genomes are compared, the functional site at position 500 is discarded since there is no exact match.
This thesis attempts to identify functional TFBSs using phylogenetic turnover and a probabilistic framework that takes binding site turnover into account. For more details, check out the PDF.
