API Reference

DecodeParams()

Constructor for DecodeParams object that sets default values for every parameter. Cost function parameters must be set by the user. If they are not, the decoding functions will throw an error.

set_xy_search_radius(prms :: DecodeParams, r :: Real)

Arguments

• prms: DecodeParams Object
• r: The lateral KDTree Search radius in pixels for aligned dots in previous barcoding blocks.

set_z_search_radius(prms :: DecodeParams, r :: Real)

Arguments

• prms: DecodeParams Object
• r: The z KDTree Search radius in slices for aligned dots in previous barcoding blocks.

set_n_allowed_drops(prms :: DecodeParams, d :: Integer)

Arguments

• prms: DecodeParams Object
• r: The number of allowed drops in a barcode. Only 0 and 1 are currently supported. The parity check matrix must have redundancy to support correcting for one drop, otherwise errors will occur.

set_lat_var_cost_coeff(prms :: DecodeParams, lvf :: Real)

Arguments

• prms: DecodeParams Object
• lvf: The lateral variance penalty coefficient in the cost function.

set_z_var_cost_coeff(prms :: DecodeParams, zvf :: Real)

Arguments

• prms: DecodeParams Object
• r: The z variance penalty coefficient in the cost function.

set_lw_var_cost_coeff(prms :: DecodeParams, lwvf :: Real)

Arguments

• prms: DecodeParams Object
• lwvf: The log weight variance penalty coefficient in the cost function.

set_s_var_cost_coeff(prms :: DecodeParams, sf :: Real)

Arguments

• prms: DecodeParams Object
• svf: The σ variance penalty coefficient in the cost function.

set_zeros_probed(prms :: DecodeParams, zp :: Bool)

Arguments

• prms: DecodeParams Object
• zp: Whether or not zeros in codewords are probed in the experiment. Drops are not supported when false.

set_erasure_penalty(prms :: DecodeParams, ep :: Real)

The penalty that each dot dropped from a barcode adds to the cost function of the corrected barcode is ep. For example a barcode with one drop adds 1*ep to the cost of the barcode.

Arguments

• prms: DecodeParams Object
• ep: penalty for decoding a barcode with a dropped dot

set_free_dot_cost(prms :: DecodeParams, fdc :: Real)

Arguments

• prms: DecodeParams Object
• fdc: The cost of not decoding a dot in the barcode candidate network

set_skip_thresh(prms :: DecodeParams, st :: Integer)

Arguments

• prms: DecodeParams Object
• st: conflicting networks of barcode candidates containing more barcode candidates than this threshold will be discarded.

set_skip_density_thresh(prms :: DecodeParams, sdt :: Real)

Arguments:

• prms: DecodeParams Object
• sdt: conflicting networks of barcode candidates that have a higher ratio of candidate barcodes to area of their bounding box in pixels than this threshold will be discarded.

get_codepaths(pnts :: DataFrame, cb_df :: DataFrame, H :: Matrix, params :: DecodeParams)

Arguments

• pnts: DataFrame of seqFISH psfs. Must include columns:

  • x : x spatial coordinate of psfs
  • y : y spatial coordinate of psfs
  • z : z spatial coordinate of psfs
  • s : the sigma width parameter of the psfs
  • w : the weight (or brightness) of the psfs

  Additionally, there the data frame must either have columns

  • block : the barcoding block in which the psf was found
  • pseudocolor : the pseudocolor of the barcoding block in which the psf was found

  or the block and pseudocolor can be computed from the hybridization

  • hyb : the hybridization in which the dot was found

  where block = ceil(hyb / q), pseudocolor = (hyb - (block - 1) * q) % q, and q is the number of pseudocolors.

• cb : The codebook.

• H : The parity check Matrix

• params : DecodeParams object holding the parameters for decoding

Computes codepaths with syndrome decoding, removes codepaths that exceed the cost of not decoding their component dots, and and returns DataFrame of candidate codepaths.

get_codepaths(pnts :: DataFrame, cb :: Matrix, H :: Matrix, params :: DecodeParams)

Computes codepaths with syndrome decoding, removes codepaths that exceed the cost of not decoding their component dots, and returns DataFrame of candidate codepaths.

Arguments

• pnts: DataFrame of seqFISH psfs. Must include columns:
  • x : x spatial coordinate of psfs
  • y : y spatial coordinate of psfs
  • z : z spatial coordinate of psfs
  • s : the sigma width parameter of the psfs
  • w : the weight (or brightness) of the psfs
• cb : The codebook.
• H : The parity check Matrix
• params : DecodeParams object holding the parameters for decoding

choose_optimal_codepaths(pnts :: DataFrame, cb_df :: DataFrame, H :: Matrix, params :: DecodeParams, cpath_df :: DataFrame, optimzer, ret_discarded :: Bool=false)

Arguments

• pnts: DataFrame of seqFISH psfs. Must include columns:
  • x : x spatial coordinate of psfs
  • y : y spatial coordinate of psfs
  • z : z spatial coordinate of psfs
  • s : the sigma width parameter of the psfs
  • w : the weight (or brightness) of the psfs
  Additionally, there the data frame must either have columns
  • block : the barcoding block in which the psf was found
  • pseudocolor : the pseudocolor of the barcoding block in which the psf was found
  or the block and pseudocolor can be computed from the hybridization
  • hyb : the hybridization in which the dot was found
  where block = ceil(hyb / q), pseudocolor = (hyb - (block - 1) * q) % q, and q is the number of pseudocolors.
• cb : The codebook.
• H : The parity check Matrix
• params : DecodeParams object holding the parameters for decoding
• cpath_df : A DataFrame output from the get_codepaths function, defined above.
• optimizer : solver for integer linear programming optimizations. A list of supported solvers is available here
• ret_discarded : if true, return data frame of candidate codepaths that were discarded for being in too large/too dense of a conflict network

Returns

• DataFrame of decoded barcodes
• If ret_discarded == True, returns second DataFrame of the codepaths discarded to density filters

Choose best codepaths from previouly found candidates that may have been found with less strict parameters. Reevaluates the costs for each candidate and trims according to the passed parameters.

decode_syndromes!(
    pnts :: DataFrame,
    cb,
    H :: Matrix,
    params :: DecodeParams
    optimizer = GLPK.Optimizer
)

Arguments

• pnts: DataFrame of seqFISH psfs. Must include columns:

  • x : x spatial coordinate of psfs
  • y : y spatial coordinate of psfs
  • z : z spatial coordinate of psfs
  • s : the sigma width parameter of the psfs
  • w : the weight (or brightness) of the psfs

  Additionally, there the data frame must either have columns

  • block : the barcoding block in which the psf was found
  • pseudocolor : the pseudocolor of the barcoding block in which the psf was found

  or the block and pseudocolor can be computed from the hybridization

  • hyb : the hybridization in which the dot was found

  where block = ceil(hyb / q), pseudocolor = (hyb - (block - 1) * q) % q, and q is the number of pseudocolors.

• cb : The codebook.

• H : The parity check Matrix

• params : DecodeParams object holding the parameters for decoding

• optimizer : solver for integer linear programming optimizations. Uses the open source GLPK optimizer by default, but allows faster commercial optimizers to be used if necessary.

A list of supported solvers is available here

Takes a DataFrame of aligned points with hyb, x, y, z, w, and s columns; codebook matix; parity check matrix (H); and DecodeParams stucture with decoding parameters set. Adds columns to input pnts matrix indicating the encoding block, syndrome component value, and decoding result indicated as the row of the codebook matrix matched to. Split up points into weakly connected components, then finds possible codeword messages and runs simulated annealing to assign them. The pnts dataframe should have hybridization, x, y, and z columns