During some time off from work this winter, I had the opportunity to learn about micro-roasting coffee from the owners of my favourite coffee micro-roaster on the Sunshine Coast, Sweet Start. They were kind enough to walk me through roasting several pounds of Ethiopian beans on their Kaldi Fortis Roaster, with a particular emphasis on the parameters required to roast Full City+ (my favourite). They had started roasting with the Whirly-Pop bbq method and worked their way up to this beautiful machine.

In addition to this experience, Kelly gifted me a Whirly Pop set up for my birthday (as well as some green beans), so I could try it myself. Today I gave it a shot and learned a number of lessons that I hope to apply with my next attempt.

I started by bringing the Whirly Pop up to 250 degrees Celsius on my BBQ propane side burner. I added 8 oz of green beans and it dropped to 164 degrees celsius before starting to rise again. I forget the temp at first crack, but I think second crack started around 190 degrees celsius. I removed the beans to cool and remove chaff at about 210 degrees celsius, which occurred at 6 minutes and 15 seconds.

A number of the stages seemed to occur too quickly at the given temp and I resulted in quite dark and oily beans. On reflection I think there are several errors I probably made, including a) starting too quickly once the thermometer reached 250 degrees celsius (might have still been heating to a higher temp), b) burner may have been turned up too high (was about midway), c) I may have not rotated the Whirly Pop quick enough leading to too much contact time.

Nonetheless, this was a lot of fun and I’m glad I have enough green beans to be able to experiment with the parameters some more. The Whirly Pop method is definitely more error-prone and finicky than the Kaldi Fortis method, but I did learn a lot about the sounds and smells of the different stages.

Those who know me know that I am obsessed with data and data visualization. I spend a good chunk of my job supporting population health surveillance projects and have developed dashboards to facilitate data-driven decision making; however, this primarily involves R, excel and to some extent tableau.

In my spare time I also dabble in computer programming. As demonstrated by this website, I have a recent interest in html5 and javascript coding. Maintaining my own website allowed me to explore javascript data visualization libraries, such as HighCharts, as well as MySQL databases. Combined, these allow amazing customization of the data visualization experience.

Perhaps one of the most famous interactive population data websites is GapMinder, developed by the incredible Dr. Hans Rosling and his team. The animations help convey the trends over time and can lead to new insights, particularly for the lay public.

I wanted to experiment with similar animations for local data, so I developed the CommunityDx Viz Lab. While currently limited to geography-specific population pyramids, I hope to use this platform to play around with different ways to visualize local data using animations. In the mean time, I hope to use these animations to demonstrate the change in age composition of populations in British Columbia. This is particularly relevant in healthcare, where many communities are seeing a considerable increase in the senior population relative to other age groups.

I recently had the opportunity to travel to the U.K. to attend and present a poster at the
Applied Bioinformatics and Public Health Microbiology Conference at the Wellcome Genome Campus, just outside of Cambridge. Although small, the conference highlighted research at the forefront of genomic epidemiology and provided insight into the future of public health microbiology.

I present the same poster from the NAR Union on our Tuberculosis WGS Report Design Study (see presentations). This was a great opportunity to get in person feedback on our findings, particularly as we are still working on the manuscript.

However, the highlights of the conference were the presentations and panel discussions. Perhaps my favourite presentation involved nextstrain.org, a web platform for visualizing the genomic epidemiology of epidemics. Not only is the website gorgeous, it presents incredibly rich information to visualize (semi) real-time evolution and
transmission within large scale epidemics. For example, the platform allows the user to visualize the evolution of the zika epidemic, from the South Pacific, to Brazil and beyond.

In addition to the conference, I had a great time exploring London. I particular, I made the pilgrimage to the former location of the Broad Street pump (as well as the adjacent and aptly named John Snow pub). While the pump is no longer there, a pink curbstone marks the location.

Although my life revolves around apple products, I sometimes have need for a windows based PC – some software is still limited to windows, including Microsoft Access and other utilities. My laptop is also limited with regards to speed and memory, resulting in difficulties processing some intensive code. I therefore decided to finally take the plunge and build a PC from individual components – a challenge I have wanted to try since I was a teenager and a core milestone in being a computer geek.

I set my budget at under $1000 – considerably less than any of my apple computers. This was higher than I initially planned; however, my initial budgeting was based on US prices, which are considerably lower than in Canada.

I purchased the following components on Amazon:

• Case – Corsair Carbide Series SPEC-01 Mid Tower Gaming Case
• Processor – Intel 3.70 GHz Core i3-6100
• Motherboard (MOBO) – ASUS PRIME B250M-A
• Graphics Card – EVGA GeForce GTX 1050 Ti GAMING, 4GB
• Memory – Crucial 8GB Kit (4GBx2) DDR4
• Hard drive – Western Digital 1 TB 3.5-Inch 7200 RPM SATA III
• Power supply – EVGA 500 W1
• Operating System – Windows 10

Given that I hadn’t built a computer from components before, I followed a step-by-step guide published by Tech Buyer’s Guru (in addition to the manuals that came with the case, motherboard and power supply).

The build process was fairly straightforward, starting with the MOBO/processor installation, RAM, power supply, hard drive and graphics card, although some of the pins on the MOBO were not labelled very clearly.

While the build seemed easy, I hit panic mode when I tried to turn it on…. the computer started but nothing displayed on the monitor. I reviewed all connections but still nothing. After troubleshooting online, I found a suggestion to remove the lithium battery from the MOBO for a couple minutes to reset. This worked and got me to the MOBO BIOS screen!

The next step involved booting from the Windows 10 USB drive. This led to my next panic moment….a windows error screen. This screen said my computer needed to be repaired and listed a 0xc0000225 error. After looking up this error online, I was horrified to see that it could result from damage to the MOBO or processor….what had I done???? It could also result from corruption of the boot files though, so I decided to create my own windows 10 USB.

In order to do this, I downloaded the Windows 10 ISO from the web and installed it to a USB key using UNETBOOTIN for mac. No luck the first time….different error (0xc000000e). I was fairly certain I broke something now; however, I wasn’t giving up. If windows didn’t work, then I would try linux. So I created an UBUNTU boot USB also using UNETBOOTIN and……it worked! I was overjoyed to see the UBUNTU boot screen (albeit I booted from the USB without installation).

Now that I knew the components were connected properly and working, I decided to try creating a bootable Windows USB again. This time I used a different USB key and ensured that the key was formatted as FAT32 prior to installing windows 10. Then came the moment of truth…..it worked!!!!!

Milestone achieved 🙂

As I finish up the fall semester, I found myself writing two papers on different aspects of the fentanyl crisis in British Columbia.

Fentanyl has been linked to the dramatic increase in overdose deaths among people who use drugs in British Columbia. According to the BC Coroners Service, illicit drug overdose deaths have increased from 273 in 2012 to 622 in 2016 as of October. Whereas in 2012, fentanyl was detected in only 5% of illicit drug overdose deaths, this has increased to 59% in 2016. This led to multiple provincial responses including the declaration of a public health emergency by the provincial health officer in April 2016.

The increasing rate of overdose deaths attributed to fentanyl is horrifying and shows no sign of slowing down. In order to visualize the evolution of this crisis, I took a break from writing papers to develop an interactive map:

This was created using the Highmaps API and represents a choropleth visualization of the annual number of illicit drug overdose deaths by Health Service Delivery Area. The data was taken from the October 2016 Illicit Drug Overdose Report by the BC Coroners Service.

Unfortunately, WordPress does not let you include javascript, so the full interactive map can be viewed here.

This summer I had the opportunity to work with Dr. Jennifer Gardy and her team on research involving tuberculosis and whole genome sequencing (part MPH practicum and part PHPM residency rotation). During this rotation I helped lead a project to develop a lab report template for clinical data resulting from a novel WGS pipeline at Public Health England. This follows from the COMPASS-TB study, which compared a WGS diagnostic pipeline for TB to the standard diagnostic approach, demonstrating that this can be done faster and more cost effectively. My interest in this project related to the difficulty with translating complicated WGS data into something that can be interpreted by clinical and public health practitioners.

ASCII John Snow

Following the first phase of the project, we created a new blog to present our results and other public health info’s projects that the team is working on. As one of the first posts, Dr. Gardy wrote a fantastic piece on the role of visualization in public health. This post really emphasizes how innovative breakthroughs (such as whole genome sequencing) aren’t useful until they can be implemented and translated into something frontline workers can interpret.

I also had fun creating the new logo – ASCII John Snow!