Concentrating Geoduck Broodstock Hemocyte DNA for WGBS

Because the minimum concentration for WGBS with Genewiz is 20ng/ul, I pooled and precipitated the DNA samples from yesterday as follows:

  • I combined tank 3 samples 15 and 16, total volume ~200uL
  • I combined tank 2 samples 025 and 026, total volume ~200uL
  • I added 140uL isopropanol and 20uL 3M Sodium Acetate pH 5.2 to each ~200uL pooled sample
  • I vortexed to mix
  • I spun at 12000 rpm for 30 min at 4C
  • I removed the supernatent and added 500uL of 70% ethanol to each
  • I spun tubes for 10 min at 12000 rpm at 4C
  • I removed the supernatent and resuspended in 20uL EB (elution buffer from the EZNA kit, which I think is 10mM Tris, pH 8.5)

Qubit concentrations: Standard 2 = 100ng/ul

Animal ID Tank Treatment Conc. (ng/ul) Total DNA (ng)
15_16 3 amb 284 5680
025_026 2 low (pH 6.8) 53 1060

We are going to send all of Tank 2 ‘025_026’ DNA and half of Tank 3 ‘15_16’ DNA, which Sam now has and planning to ship on Monday. The remaining Tank 3 ‘15_16’ DNA is in my box labeled ‘Shelly Trigg start date 10/11/2018’ in the -20C in rm 213.

Oyster seed proteomics time x temperature

Creating differential network visualizations

To create differential network visualizations, we need:

  1. protein relationships
    • this could be GO annotations, KEGG pathways, or even protein interaction information (although I’m not sure how conserved this is between species)
  2. a metric to quantify abundance changes
    • nodes could be colored by fold change and/or effect size to show up or down
    • node size could be relative to p-value, or even PCA loadings values
Calculating log fold change and p-values

Emma suggested calculating the ratio of ‘NUMSPECSTOT’ of each protein to the sum of NUMSPECSTOT for the sample. Then calculating log fold change. And using a proportions test for calculating the p-value.

I think all fold changes should be relative to day zero so that we can visualize what protein networks are active at each stage of development and compare differences between the two temperatures (sort of like what these guys did). For a figure, what I’m picturing is a 2 x 6 grid for the two different temperatures and 6 different days. And a network in each box of the grid showing proteins colored differently based on their fold change and sizes based on p-value significance.

Here is the R markdown file I made for calculating these. I used a Chi square proportions test for the p-values. So far I have only done one comparison (23C_day 3 to day 0).

Need to:

  • make a loop to do the rest of the comparisons

  • map the proteins to Uniprot accessions using an updated uniprot DB.

  • figure out what proteins are connected either by GO or physical interaction information