Russell's Blog

New. Improved. Stays crunchy in milk.


Posted by Russell on October 24, 2009 at 5:58 p.m.
Looks like I got my gene to grow in E. coli!

The colonies that didn't get the plasmid I'm using to carry MXAN7396 turn blue when grown with X-gal (bromo-chloro-indolyl-galactopyranoside). The ones that got the plasmid don't.



Posted by Russell on October 21, 2009 at 4:04 a.m.
I finally got through the double-PCR phase of the protocol without wrecking something. Yay!

My hacked up version of the gene gene is getting snipped up with everyone's favorite restriction enzymes (BamH1 and EcoR1). Then I get to splice it into a plasmid, and electroport the plasmids into some cells, and maybe they will do something interesing.

The cloning blues

Posted by Russell on October 19, 2009 at 8:43 p.m.
I've been doing my first laboratory rotation in Mitch Singer's lab, and trying to learn what people are actually doing when they publish these spiffy experimental results. So far, I've mostly been wrecking things. Fun disasters of the week :
  • Wrecked a DNA extraction by grabbing the wrong Pipetter and putting 300 microliters into a tube instead of 3.
  • Misread an illegible label and used butanol instead of ethanol, destroyed second attempt at the aforementioned DNA extraction.
  • Dropped the wrong tube in the trash, screwed up the third attempt at the aforementioned DNA extraction.
  • Kept a gel on the UV bench too long while trying to chop out little cubes with a razor blade, annihilated all the DNA, and screwed up fourth attempt at aforementioned DNA extraction.
  • The PCR cycler didn't close correctly, and my reaction tubes evaporated; screwing up fifth attempt at aforementioned DNA extraction. (At least this one wasn't my fault.)
I'm now spending the evening in the lab running everything over again, for the sixth time. Yays!

I definitely sticking to informatics -- that part of the rotation is going pretty well. I'm just not cut out for benchwork.

SmartMeter data from PG&E

Posted by Russell on October 05, 2009 at 1 a.m.
PG&E is no angel, but it deserves credit for making what looks like a good-faith effort to get California off of carbon. It's serious enough to have quit the US Chamber of Commerce to distance itself from the group's extremist views on climate change.

PG&E still owns six coal burning power plants, curiously located in Florida, New Jersey and Pennsylvania (presumably it uses them to swap power with other generators). It generates about 46% of its electricity from hydroelectric dams.

Rucker Creek dam, a small PG&E facility in Nevada County

One of the more interesting projects PG&E is undertaking is improving the resolution of its demand monitoring using SmartMeters. There is a lot of hype about the "Smart Grid," but basically it boils down to realtime use monitors, like these :

that are wired up to report the data somewhere. It's basically an off-the-shelf Tweet-A-Watt.

According to the PG&E web site, they are using SmartSynch meters, which use TCP/IP over some kind of wireless network. It's difficult to find information about the hardware itself, probably on account of the assorted idiots wetting their pants about people h4X0ring their refrigerators (actually, I don't know if Bill Mullins is an idiot, but his article about smart meters is depressingly typical).

Yes, it is possible for a bad person to break into your PG&E account to obtain this data.1 But so what? Power meters are inductively coupled to the circuit they measure. They can look, but they cannot touch. IOActive, a security research firm, claims that they can break into certain smart meters and "cut off power." I suppose we are meant to construe this as "cut off power to the house," but that isn't what power meters do. That is what those huge knife switches, with the lock-out-tag-out rings, are for. I'm skeptical that a certified electrician would work on a residential circuit with a computer controlled on-off switch. I certainly wouldn't. What "cut off power" probably means is that they can shut down the microcontroller, and stop the meter from collecting or reporting data. We're left to speculate, though, because the report is confidential. I speculate that they are hyping a buffer overflow exploit to gain as much attention as possible.

Nobody is going to h4x0r your refrigerator and reprogram it to be an E. coli chemostat. If you are worried about your personal data floating around on the big bad internets, your worries are better directed at your bank and your health insurance provider. The bad guys don't care that you left your bathroom light on all night last Thursday; they just want the routing number for your savings account.

While the data isn't very valuable for nefarious purposes, it is extremely valuable in the noble (if mundane) pursuit of frugality. Here's what PG&E shows you if you've been upgraded to a smart meter :

Having the graphs is neat, but the usability of the site is poor. Fortunately, they let you download the data as CSV files, although you have to go a week at a time. It's all very 1995. Happily, is working on a real-time data browser tool called Power Meter which will make this a lot nicer. For now, I just wish I had an XML-RPC interface.

I've already learned something from this data. On the 29th and 30th, I was at the Granlibakken conference center for the UC Davis Host Microbe Interaction conference. Those days show dramatically less power use between about 22:00 and 2:00, which is when I'm usually hacking at my desktop machine. One more reason to start thinking about replacing this behemoth.

1. Actually, it's stupidly easy to gain access to someone's PG&E account if you have their account number. Just create a new web account, type in the account number, and there you go! Now you can really fuck with them by paying their bill, which is about all you can do with a PG&E account.