Matt:LabNotes/2015-11-15

From ZhangLabWiki
Jump to navigation Jump to search

Analysis of DARTFISH Tile Image to compare with RNAscope[edit]

Running Matlab 2D Decoding[edit]

  • Used 2D Decoding
    • Program folder: C:\Users\Matt\Dropbox\GradZhangLab\Matlab\DARTFISH_PxPx_Organized\
    • Data folder: C:\Users\Matt\Documents\RolonyImages\11-11-2015\20151029_BA8_DARTFISHsuppv2_20XDecoding\
  • First time trying on images taken with 20X objective (normally 63X) but kept the 2048x2048 pixel resolution
    • Each image went from 184um to 584um width
    • Pixel size went from 0.09um to 0.3um
    • Axial resolution went from ~0.5um to ~2um
    • Z step size went from 0.3um to 1.04um
  • DecodePxPx.m paramaters set in Info.m file
    • Area Lower = 2 pixels
    • Area Upper = 15 pixels
    • TreshDistance = 0.05 (normalized intensity below which to call '0')
  • Alignment Offset (iCycle 2 as reference) Results:
    • Offset, Pos 1, iCycle 2 & 1, [x, y] = [1, 1]
    • Offset, Pos 1, iCycle 2 & 3, [x, y] = [0, -1]
    • Offset, Pos 1, iCycle 2 & 4, [x, y] = [1, -2]
    • Offset, Pos 1, iCycle 2 & 5, [x, y] = [0, -2]
    • Offset, Pos 2, iCycle 2 & 1, [x, y] = [1, 0]
    • Offset, Pos 2, iCycle 2 & 3, [x, y] = [1, 0]
    • Offset, Pos 2, iCycle 2 & 4, [x, y] = [1, -2]
    • Offset, Pos 2, iCycle 2 & 5, [x, y] = [0, -2]
    • Offset, Pos 3, iCycle 2 & 1, [x, y] = [1, 0]
    • Offset, Pos 3, iCycle 2 & 3, [x, y] = [1, -1]
    • Offset, Pos 3, iCycle 2 & 4, [x, y] = [1, -2]
    • Offset, Pos 3, iCycle 2 & 5, [x, y] = [0, -2]
    • Offset, Pos 4, iCycle 2 & 1, [x, y] = [1, 1]
    • Offset, Pos 4, iCycle 2 & 3, [x, y] = [0, -1]
    • Offset, Pos 4, iCycle 2 & 4, [x, y] = [0, -2]
    • Offset, Pos 4, iCycle 2 & 5, [x, y] = [0, -1]
    • Offset, Pos 5, iCycle 2 & 1, [x, y] = [1, 1]
    • Offset, Pos 5, iCycle 2 & 3, [x, y] = [0, -1]
    • Offset, Pos 5, iCycle 2 & 4, [x, y] = [0, -2]
    • Offset, Pos 5, iCycle 2 & 5, [x, y] = [0, -1]
    • Offset, Pos 6, iCycle 2 & 1, [x, y] = [0, 1]
    • Offset, Pos 6, iCycle 2 & 3, [x, y] = [0, -1]
    • Offset, Pos 6, iCycle 2 & 4, [x, y] = [0, -1]
    • Offset, Pos 6, iCycle 2 & 5, [x, y] = [0, -1]
    • Offset, Pos 7, iCycle 2 & 1, [x, y] = [0, 1]
    • Offset, Pos 7, iCycle 2 & 3, [x, y] = [0, 0]
    • Offset, Pos 7, iCycle 2 & 4, [x, y] = [0, -1]
    • Offset, Pos 7, iCycle 2 & 5, [x, y] = [0, -1]
    • Offset, Pos 8, iCycle 2 & 1, [x, y] = [0, 0]
    • Offset, Pos 8, iCycle 2 & 3, [x, y] = [0, -1]
    • Offset, Pos 8, iCycle 2 & 4, [x, y] = [0, -1]
    • Offset, Pos 8, iCycle 2 & 5, [x, y] = [0, -1]
    • Offset, Pos 9, iCycle 2 & 1, [x, y] = [0, 1]
    • Offset, Pos 9, iCycle 2 & 3, [x, y] = [0, 0]
    • Offset, Pos 9, iCycle 2 & 4, [x, y] = [0, -1]
    • Offset, Pos 9, iCycle 2 & 5, [x, y] = [0, 0]

Stitching Results Into Tile[edit]

  • DecodePxPx.m results containing coordinate info is stored in StatTable matrix saved in 'Analysis_BA8_suppv2_2015-10-29_PosX_Try_1.mat'
    • Column order: Index, Area, Avg Intensity, X Coordinate, Y Coordinate, Total Count
  • Output following info into one consolidated excel file:
    • Column order: GeneName, GeneBarcode, Area, X Coordinate, Y Coordinate
    • Use script RolonyPositionExcels.m
  • Aligned/Cropped image sizes
    • Pos1: 2044
    • Pos2: 2046
    • Pos3: 2046
    • Pos4: 2044
    • Pos5: 2044
    • Pos6: 2046
    • Pos7: 2046
    • Pos8: 2046
    • Pos9: 2046
  • Finding size of overlap between each position
    • Based on stage coordinates expect overlap size to be 295px = 84um
    • To find actual overlap compare positions of decoded rolonies
      • Did this manually so only looked at most abundant gene 'SNAP25'
      • In the same region, more rolonies decoded at the bottom of previous image then top of latter image
        • E.g. More SNAP25 in bottom 279px of Pos1 than in top 279px of Pos2
        • Could be due to photobleaching? "uneven illumination" or equivalent phenomena for confocal?
  • Offsets (calculated average Y-offset of SNAP25 rolonies that overlapped)
    • Pos1->2: 279
    • Pos2->3: 289
    • Pos3->4: 298
    • Pos4->5: 278
    • Pos5->6: 278
    • Pos6->7: 305
    • Pos7->8: 280
    • Pos8->9: 299
  • Stitched Tile A:
    • Use count of previous position for overlapping region
  • Stitched Tile B:
    • Use count of latter position for overlapping region

Counts of Genes in 21 Y-slices[edit]

  • Counted number of in bins of 752 pixels

Counts of Nuclei in 21 Y-slices[edit]

  • Since Positions 1-4 of 9 during imaging had weird nuclei morphology I took altPositions directly adjacent (no shift in y direction) and treat those as if
  • Due to these altPositions I can't make one long tile image so will just count nuclei in the appropriate regions
    • eg Y-slice 1 = 1-752 of Pos 1
    • eg Y-slice 3 = 1505-2048 of altPos1 + 280-489 of altPos2
  • Use CountNucleiPerYSlicesfromMIP.m which calls nucC.m
    • strelSize = 10

Number of Rolonies per Nuclei[edit]

  • Results are in C:\Users\Matt\Documents\RolonyImages\11-11-2015\20151029_BA8_DARTFISHsuppv2_20XDecoding\Aligned\BA8_suppv2_2015-10-29\Excels
    • Update wiki later with graphs
  • Note: 20X Decoding (with 0.75 NA) is not good enough
    • In 9 positions equaling 2.677mm^2
      • Decoded 6,908 rolonies (and this includes double counting rolonies of overlapping regions)
      • 225 genes
      • 178 genes > 3 counts
    • In 4 positions of 63X decoding equaling 0.1354mm^2
      • Decoded 1,641 rolonies using 3D Decoding (~10% more than 2D Decoding)
      • That would be >30,000 rolonies in 2.677mm^2
  • Maybe try a 40X 1.3NA objective?
    • Especially for mouse embryo project
      • 292um FOV should be enough to fit each embryo section
      • 1.3NA should be high enough resolution for 3D Decoding