Author Archives: Rachel Hertog

Can oysters filter laby? Experiments and Models

Our group is focusing our project on the question of whether mitigation of laby outbreaks in eelgrass beds is possible via two methods, culling of disease or filtration by oysters.

Today we set up round two our oyster filtration experiments again in Lab 5. Our hypothesis is that oysters will filter laby out of the water, reducing the prevalence and/or severity of infection in exposed eelgrass.

cloudy_laby_setup2

I’ve been analyzing the data from our first round of experiments and the results are interesting. There is no difference in prevalence of infection between the eelgrass that was in the seawater+laby treatment versus the seawater+laby+oyster treatment (though both of those treatments had significantly higher prevalence that the treatments that we did not add laby to, so it appears we can infect eelgrass via exposure, which is pretty neat). There also does not seem to be a change in the severity of the infection (measured as total blade length/total lesion length) in the seawater+laby compared to the seawater+laby+oyster treatment. I’m interested to see if our results are the same the second time around.

Before we worked on our experiment, we spent some time with Maya learning about disease modelling and had a chance to play around with some software and a model she developed for oyster filtration of a pathogen. The first 20 or so times I ran the model the pathogen in the water was eliminated, but I wanted to see if I could get it to be filtered, but still present in the environment. It took some tinkering with the parameters, but I finally got the pathogen to persist at low levels. http://youtu.be/tT9Waxrwstk

Fisherman Bay eelgrass

**the images in this post have been updated to reflect the correct station numbers

I spent part of my day working with the data from the eelgrass collected from Fisherman Bay on 8/9/14.

First, I looked at prevalence, or the proportion of infected blades. I looked at the entire bay, and then broke down prevalence by station.
fisherman_bay_prevalence_station

This is pretty cool as it shows a distinct change in prevalence from the mouth of the bay (station 1) to the stations further in. Sandy gave me this map today showing the 5 stations, 4 of which had eelgrass and were sampled last friday.
SampledWithSeagrass_labeled

Next, I looked at prevalence by size class. As we all noticed, these blades were much larger than the ones we sampled from the intertidal at Indian Cove and False Bay.
fisherman_bay_size_class
Finally I took a stab at severity of infection by graphing blade length vs. total lesion length.
fisherman_bay_length_plot
I think understanding this relationship between blade size, site and infection will take some more complex math than had time for today. Looking forward to Maya’s talk on Thursday morning for more insight into how to approach this data.

BinGO was his name-o

I spent a good amount of time today trying to get the BinGO plugin for Cytoscape to work, only to come to the conclusion that it may be a waste of time. Since ReviGO produces xgmml files that can be read by Cytoscape directly, BinGO would just be replacing the David to ReviGO step, which so far seems to work pretty smoothly.

Unfortunately, I spent so much time wrestling with reformatting files and perusing help groups I didn’t make any new pretty pictures, so my pictures from yesterday will have to suffice.

Until tomorrow.

Data Visualization

Today was another challenging day working with our transcriptome data. Ending the day with visualizations of our data was a really useful way for me to get a better sense of how to understand and interpret our data. It also is really fun. I’m interested in learning more about what makes a good data visualization and learning some new tools to produce them.

There are more details in my IPython notebook about exactly what I did today and what programs I used to get from data to visualization. Here’s one of the figures I made with Revigo. I picked it not because it’s the best visualization, but because I’d like to explore this kind of network visualization more in the coming week to figure out how I can use it for myself and to communicate what we’re learning with other people.

revigo_interactive

Eelgrass experiments, field and lab

~I seem to have neglected to post this on Saturday, so today is will be a twofer from me. I’ve added some more information to the bottom to reflect what I did this afternoon~

The other day I posted a few graphs of the infection prevalence in eelgrass from Indian Cove and False Bay. The differences between the control and culled plots were hard to eyeball, so I decided to dig a little deeper into our data. At Indian cove, 36% of the blades in the control plots had one or more lesions, while just 27% of blades in the culled plot had lesions. We got similar results in False Bay where again, 36% of blades in the control plot had lesions and 29% of the blades in the culled plots had one or more lesions.
(Fisher’s exact test, p>0.05 for both sites. However, the pvalue for Indian Cove was p=0.055).

At Indian Cove, there was a significant difference in blade area between blades with and without lesions at Indian Cove (p0.05), but there was a significant difference (p<0.05) in the culled plot. Again, the lesioned blades were larger than those without lesions.

In the Groner et al (2014) paper they discuss a much more sophisticated model for exploring this kind of data, so I'm hoping to get a chance to sit down with Maya sometime this week and see what else we can learn about this experiment.

***
For our laby experiment in the lab we have been taking pictures everyday of each of the young eelgrass in each beaker in order to track the development of lesions over time.

Lesion photo day 1

Today was my fourth day photographing the blades, and the first day that I have tried to quantify the changes in number and shape of the lesions. I’m using a software package called ImageJ to circle the lesions on each blade and measure their dimensions. The image analysis has been a bit harder than I expected because the leaves have darkened significantly in the last four days so seeing the lesions on the blades is a challenge. Also, some of the leaves are coming off of their rhizome so the plants look different than they did on the first day, making it difficult in some cases to figure out which blade goes with which plant. I’ll have more to say about this tomorrow as I move through the rest of the photos.

I love terminal and terminal loves me

Today we dove headfirst into terminal, blast and (my favorite of the day) IPython. I’m relatively comfortable with working in a terminal, but everything else was all new to me. I’m especially excited about the learning more about how to use IPython. Lately I’ve been thinking a lot about making a more complete transition from my paper notebooks to a digital notebook but have had a hard time finding a program that worked for me.

In the afternoon Steve gave us some context for who he is and what his research interests are. He’s a strong advocate for practicing open science, which includes having digital notebooks, sharing code and data and maintaining an online presence. One issue that came up today, and has come up in every conversation I have ever had about open science, is the fear of being scooped or having your data “stolen”. This question is so common that the open science group at UC Davis dedicated a whole workshop to the issue. Because they’re an open science group, the whole workshop called “Getting famous or getting scooped? Risks and opportunities in sharing your science and scholarship” is on youtube at

It’s a bit long but was an interesting conversation about how open science works and why we should all be doing it. Today was a good reminder that I need to be more proactive about exploring my options around not just publishing my work in journals but getting it out via other avenues.

Preliminary eelgrass data

Our preliminary eelgrass data from the Indian Cove and False Bay culled plots are available on the google docs spreadsheet.   Tonight I made a couple of quick graphs looking at prevalence of infection comparing the culled vs. control plots at both sites.

Prevalence of lesions on eelgrass in False Bay

Prevalence of lesions on eelgrass in False Bay

Prevalence of lesions on eelgrass in Indian Cove

Prevalence of lesions on eelgrass in Indian Cove

It looks like the proportion of infected individuals is lower in the culled plots at both sites, but I’m not sure yet if the difference is statistically different yet. I have some ideas about other questions to ask with our preliminary data, but that number crunching will have to wait until tomorrow.