Daniel:Notebook/PosSequencing/2017-4-14

From ZhangLabWiki
Jump to navigation Jump to search

Bead Loading Analysis (Started Tuesday)[edit]

Back to Calendar

Image Analysis (Started Wednesday)[edit]

Analysis of the images taken on Wednesday. Since there are only 6 and I'm not planning on doing this long, I did the analysis manually. Here's the pipeline, done in Powerpoint.

  1. Start with Dapi image and identify (by eye) the bright clusters; circled in red
  2. File:20170412-D1Emulsion-DAPI3-Analysis.png
  3. Use "change image function" to replace DAPI image with brightfield image. The change image function keeps the position and dimensions of the original
  4. Mark and count (in blue) the beads that appear in the same plane as the red circled dots but had no signal. These are confirmed no-calls. The green circled dots are "wild cards", and are possibly no-calls, although they might just be out of phase which lowers there signal
  5. File:20170412-D1Emulsion-BF3-Analysis.png
  6. With the counts, I calculated the percentage of empties. Since I can't tell whether any signal is a single or double, I only count empties. This fraction, I assume is the Poissonian result of loading. Therefore, using the Poisson distribution, I can estimate the average loading, and use the Poisson CDF to estimate the number of beads with 2 or more (1-P(2)). The results are recorded below.
    Field # Dapi Stained Droplets # Non-Dapi Droplets Wild Cards (Droplet?) Loading Percentage Loading Pct with Wildcards Likely Doubles % Likely Doubles % (+WC)
    1 8 9 14 47.1 25.8 2.7 0.4
    2 5 4 9 55.6 27.8 4.9 0.5
    3 10 9 25 52.6 22.7 4.0 0.2
    4 11 5 23 68.8 28.2 11.3 0.5
    5 12 35 10 25.5 21.1 0.3 0.2
    6 22 15 24 59.5 36.1 6.3 1.1
    Total 68 77 105 46.9 27.2 2.6 0.4