Comments for Neuromancy Blog

Comment on Optogenetics in 1000 words or less by todd

todd — Tue, 19 Apr 2011 05:25:51 +0000

Nice summary. Also worth noting that light stimulation using photosensitive neurotransmitters and the same type of fiber optic probes was accomplished in the 90s. Also Hegemann deserves credit for isolating the proteins to begin with. Great technique!

Comment on Quickie – Considerations of in vivo electrophysiology by Neuromancy

Neuromancy — Tue, 21 Dec 2010 15:26:29 +0000

Sorry? I don’t understand.

Comment on Quickie – Considerations of in vivo electrophysiology by rresalrie

rresalrie — Tue, 21 Dec 2010 06:07:10 +0000

“Jacobs notes “this may change, of course, as more detailed analyses are carried out.” The paper was published in 1986, and for my neuron of interest – dopaminergic neurons – this has indeed changed.”
How much is real?

Comment on I’m learning about bees! by Mike Lisieski

Mike Lisieski — Wed, 10 Nov 2010 18:03:40 +0000

Yay for honeybee neuroscience! I learned about the neuron VUMmx1 a few months ago, and was absolutely amazed that associated learning depended so critically on a single neuron (although, I guess I shouldn’t be.)

Best of luck, and esp. with the whole getting-kids-and-teachers-concerned-about-ecology – it’s a truly noble cause.

Comment on Scatterplot matrices and MANOVA in R by Neuromancy

Neuromancy — Thu, 04 Nov 2010 14:51:37 +0000

Well I’ve done separate ANOVAs of each DV, and the most significant one is p=0.1034, so I don’t think I need to bother with a MANOVA even if I could find a statistic where my data satisfies the assumptions do I?

Cheers for all your help Tim!

Comment on Scatterplot matrices and MANOVA in R by Neuromancy

Neuromancy — Thu, 04 Nov 2010 14:10:44 +0000

Yeah, I was hoping to have two regressions for the cell property data where there are two groups, and just one group for the response properties where there are just the excited cells. I guessed that I might have to construct the plot manually for that though. Have you got any experience using the ggplot2 package? It seemed like it might be easier for visualisations, I’m pretty sure it can do multiple plots like this. The q-q plots aren’t a big deal, I just thought they’d be the most useful at the moment.

I just tried doing the MANOVA this morning, and it threw an error “Error in summary.manova(fit, test = “Wilks”) :
residuals have rank 6 < 7" which I think means I have too many DVs compared to rows in each group? I'm not sure whether I can combine the light response data and the TNC stim data into one MANOVA, as most of the rows in TNC are data points from the same cell as rows in the light flash data.

Comment on Scatterplot matrices and MANOVA in R by Tim Lucas

Tim Lucas — Thu, 04 Nov 2010 13:59:05 +0000

I’m not quite sure what you want to do with the light data. Do you want to group it by Responsive, unresponsive? That way on some plots you’ll have 2 data sets and 2 regressions (one for excited and one for unresponsive) and when there is no data unresponsive you’ll just have 1 regression?

If so I don’t think you can do it with spm() because you need to change how it groups the data (from 2 groups to 1) and what data to use. You’ll just have to do it manually. I couldn’t get the R^2 values out either, so to put them on you’d have to do all the regressions seperately anyway. It won’t be too many lines of code to do all the regressions and then plot them all with layout() or something. Might be a bit more of a pain getting the q-q plots in, but shouldn’t be too bad.

Comment on Scatterplot matrices and MANOVA in R by Tim Lucas

Tim Lucas — Thu, 04 Nov 2010 00:28:18 +0000

I’ll read this properly in the morning and post proper answers to things. To add R^2 you add them manually with text(). You save your regression to a variable, then plot the R^2 part of the regression object. In this example you would save your spm object. Then somewhere saved in that object should be the R^2 for the regressions. (I’ll find it tomorrow once I’ve installed R on kats computer. If you want to write R^2 = 0.33 and make it look nice you’ll have to round() the value and then use expression() and paste() to format it properly.

Basics of MANOVA are here:
http://www.statmethods.net/stats/anova.html

but as you said, if the assumptions are invalid then there’s no point.

I’ll actually copy and paste your script tomorrow and try and help with your main problem.

Comment on Looking back on my PhD by Viara Mileva

Viara Mileva — Tue, 02 Nov 2010 00:27:30 +0000

Don’t envy us, long-term-PhD. – it’s not that much better I’m from Canada, and seem to have taken quite the opposite road to yours – started in Biology, did Neuroscience/Electrophys M.Sc. and then switched into a psych Ph.D. I’m nearly done (in less than 5 years!! but 7 if you count the Master’s…) but I can tell you, no matter how long it takes, you NEVER feel like an expert, particularly standing up there in front of everyone at conferences, in front of the *real* experts… I wish you luck!

http://www.dreamsoffulfillment.com
http://www.naivejar.org

Comment on “Upstream” sci journalism – Public communication of science in progress by Neuromancy

Neuromancy — Thu, 28 Oct 2010 14:01:24 +0000

I guess reporting what’s going upstream in the research process isn’t going to be too interesting to the general public for the most part, but as you say Mike, it’s nice to have that record for other researchers to read, and maybe for non-scientistis to dip into.

I thought upstream communication would be better than a media summary of the end result, but improving the focus of the way in which the end result is communicated is probably a more productive, albeit possiblya less achievable goal. Unfortunately, university press offices are sometimes as guilty of sensationalising research finidngs as anyone. In these cases, however, the crime of the media is then not asking questions.