Grace’s Notebook: Friday, November 17, 2017

(1) Worked on updating the SOP a little bit more today, but I don’t think I really contributed much. I focused mainly on the first two sections. I read through the last two sections, and I think I’ll understand them better if I can work through it once we re-calibrate the probe. I think working through it will allow me to pinpoint which pieces of information I should add in order to make the process smooth in the future.

(2) I re-took some images from Pre-OA C. gigas histology slides. I was annoyed at myself for not mapping out the slides when I imaged them originally, so I did that today for future reference:


I re-imaged: 01, 04, 05, 07, 08, 12, 15, 18, and 20. I was able to add some new information to the classification spreadsheet I’m working on based on these new images.

I am having an exceptionally difficult time figuring out what Gigas-01 is. Here are some images I took of the only non-connective tissue in the sample:

Magnification: From right: 40x, 40x, 10x, and 4x.

(3) I received some more reading material for the Crab project in Juneau from Greg Jensen.


Grace’s Notebook: Thursday, November 16, 2017

(1) Worked more on updating the SOP. Still not nearly clear enough. Have to add more specifics.

(2) Worked more on the C.gigas histology classification.

Tomorrow I’ll retake some images because I think I took pictures of connective tissue rather than gonad. It’s harder to tell what is gonad when they are in the early stages of development.

I am comparing the gonad cells to those in these images from Fabioux et al. 2005:

Here is what I’ve done so far:


(3) Finally, I’ve been emailing with Pam Jensen about the crab work in Juneau coming up. She has provided more information on how that week will be laid out, as well as some references and papers to read. I have begun reading some today.

Grace’s Notebook: Wednesday, November 15, 2017

(1) I worked more on classifying Yaamini’s C. gigas gonad histology slide images. Here is the spreadsheet of the attempt at classifying I did today.

(2) Working on updating the titrator SOP from Hollie so that it reflects what our specific process will be.

Future work:

Dan dropped off the remaining OA Larval samples today, so when I have time, I’ll start imaging those. There’s quite a bit more tubes than I was expecting.

The priority is getting the Titrator going, so once the Temperature probe arrives, we’ll re-calibrate the pH probe and then start moving through the samples!

Yaamini’s Notebook: Environmental Data Meeting Part 2

Some more information from Micah and Alex

Micah send us environmental data with the following information:

These are roughly 200,000 observations at 10-minute intervals spanning eelgrass and bare at our five sites from Jun 1 to July 20ish, 2016.

Key: WB/SK/PG/CI/FB = Willapa Bay, Skokomish, Port Gamble Bay, Case Inlet, and Fidalgo Bay E/B = Eelgrass, Bare pH, pHT, do, doT = pH (total scale, calculated for in-situ temperature), temperature recorded by pH sensor, do (in mg/L), temperature recorded by do sensor (XX) = sensor label, just ignore this Some quick notes: We have no pH data for PGE, because both sensors failed. Also no data for SKB and SKE before end of June. Sensors were first deployed at that site on 6/22/16. There are no data of any kind before that date. Both dissolved oxygen sensors at FB gave some very high readings, and the DO sensor at FBE appears to have failed some time in July (yielding negative and sky-high values). I left these in to provide the data in its rawest form. Many of these sensors were exposed on extreme low tides, and we may want to clip out those data.

And Alex sent some C. gigas physiological information as well:

There are some gaps that are easy to address: i need to reweigh some vials to get tissue weight, we need to match proteome vials with isotope vials, etc.. and some not so easy to address: we crushed 10 of 15 shells per habitat in each site – in an embarassing lack of foresight we did not check that these were the same oysters used in biomarker analyses… these can be crushed but will take a while.

Before our meeting, Laura visualized the environmental data. Willapa Bay had hotter temperatures than the other Puget Sound sites, which could play into my resutls.

Here are some notes from our meeting:

  • Pulling out air time/times the sensors were exposed to the air
    • Keep air time in for temperatures, since the oysters would be exposed to the same temperatures
    • Need to cross reference tidal data for pH and dissolved oxygen measurements
      • We can pick out tidal heights and times for those thresholds, then use it to clip out data where the pH and dissolved oxygen temperatures were exposed
  • Consolidating environmental data for bare and eelgrass habitats
    • In environmental data section, can show different and significant results
    • Can see if variation in bare and eelgrass measurements led to variation in protein abundances for each site, especially with pH
      • Can easily visualize bare and eelgrass datapoints in boxplots using jitter
    • Will need to consolidate data somehow, but an option was never given…
  • Uses dissolved oxygen sensor for temperature measurements
  • Could do boxplots on a weekly basis
    • Capture week effect of pH, DO and temperature variation
  • Daily trends
    • Micah suggested picking a window of time for max and min pH
      • 3 a.m. to 5 a.m. for pH min, 4 p.m. to 6 p.m. (or 6 p.m. to 8 p.m.) for pH max
      • Windows good for caculating means, understanding significance
    • Average daily trend plots in R
    • Quantify temporal scale of variability
      • Acute exposure times
        • Plot frequency of extremes
        • Plot total time in “extreme” condition based on thresholds
  • Annectodal site information
    • Case Inlet and Fidalgo Bay never fully exposed at low tide
    • Port Gamble Bay, Skokomish River Delta, and Willapa Bay fully exposed at low tide
  • Information we will get soon
    • Micah
      • Date and time of sampling
      • Chlorophyll and conductivity information
      • Methods and R code for clipping out air exposure information in pH and DO data
    • Alex
      • Sent information on C. gigas samples I used for proteomics, will provide all correlated information
      • Will try and track down family information for outplanted siblings
    • Emma
      • Should have information on the samples I’m rerunning tonight, will walk through samples in Skyline for me while I’m gone
      • Rerunning technical duplicates will be $2000, triplicates will be $2800


Laura’s Notebook: Initial look at Environmental Data

Received environmental data from the 2016 DNR outplant today. Plotted and ran some stats in thie R Script

First, I plotted using Plotly. Note: I haven’t figured out how to embed Plotly plots in my GitHub posts yet, so these are screen shots of the plots. To view, download the .html files and drag them into your brower.

pH Continuous Data

pH Continuous Plot Screen Shot

pH Box Plot Data

pH Box Plot screenshot

DO Continuous Data

DO Continuous screenshot

DO Box Plot Data

DO Box Plot screenshot

T Continuous Data

T Continuous Data screenshot

T Box Plot Data

T Box Plot screen shot

Then I checked out normality. Temperature data was not normal, so I double-log transformed. NOTE: haven’t thrown out any outliers yet; I may after discussing data with Micah. Resulting qqplots & histograms:

Environmental Data Normality Check

Then, I ran the Bartlett Test of Equal Variance for each environmental variable, by each grouping factor.

Treatment Habitat Site Region
pH-F 9177.01824 518.84135 4287.46341 3046.27905
pH-P 0.00000 0.00000 0.00000 0.00000
DO-F 6629.77718 114.20095 5479.65027 4290.31405
DO-P 0.00000 0.00000 0.00000 0.00000
T-F 7745.15337 7.18461 7666.51686 2512.32029
T-P 0.00000 0.00735 0.00000 0.00000

Finally, I ran 1-Way ANOVA tests for each environmental variable, by each grouping factor:

Treatment Habitat Site Region
pH-F 5756.8843 18550.3358 4435.2426 5178.1590
pH-P 0.0000 0.0000 0.0000 0.0000
DO-F 462.7035 172.7403 674.5283 675.6917
DO-P 0.0000 0.0000 0.0000 0.0000
T-F 3327.6405 53.9099 7441.9589 10106.3882
T-P 0.0000 0.0000 0.0000 0.0000

from LabNotebook

Grace’s Notebook: Tuesday, November 14, 2017

Scallop Gonad Histology: I organized all the cassettes because they were many samples that we didn’t need mixed in tubs with samples we do need. I sent out the gonad tissue samples that Katherine specified along with two extras (241 and 361) because there was not a good place to put them.

C. gigas gonad classification: I made a spreadsheet to organize the classification process. So far I’ve only determined the sex of about half of the samples. The classification is going to be more challenging because I haven’t been able to find a good paper detailing the classification of female C. gigas. And most of them so far seem to be females – at least the ones that are easier to sex.

Imaging Oly larvae: I spent some time on this today. However, I found out that all the tubes that I have don’t actually need images taken, because they already have data. I believe Dan will bring over some sample tubes for ones that need images taken sometime this week and then I’ll get that all finished up.

Titrator: The Mettler-Toledo buffers arrived today! So we can recalibrate the pH probe. I also purchased a Temperature probe, since we don’t have one.

Yaamini’s Notebook: Remaining Analyses Part 6

Looked into ROC curves

Receiver operating characteristic curves, or ROC curves, can be used with proteomic data to visualize how sensitive and specific a potential biomarker is to the environment. I wanted to make some ROC curves for peptides that were differentially expressed between Puget Sound sites and Willapa Bay. In addition to this paper on ROC analysis, I found some resources on how to make ROC curves in R. It seems like I have a few options:

  1. pROC
  2. simple.ROC
  3. roc.plot
  4. Base R

I haven’t really looked into the differences between these methods, but what I can tell is that I’ll need to first build a model with my data. The simplest model I can build is something like the following:

protein abundance ~ site

However, I think my model will be more powerful (and the ROC curves more useful) if I include environmental variables and Alex’s data as covariates:

protein abundance ~ site + habitat + pH + temperature + DO + conductivity + chlorophyll + tidal depth + %c + %n + tissue mass + shell strength + shell density

Obviously, I can use covariate deletion to identify the most significant model before I proceed, or I can only include covariates that contribute to the differential protein expression. I think I’ll also need to use a binomial model with a logit link, since all of the resources mention that the ROC analysis is evaluating successes and failures.

Bottom line: I need to wait until I get the environmental data before I can proceed.