NEDA employs an input configuration file in text format to provide input/output paths and parameters.
For this input configuration file, we provide:
Essential and Auxiliary Parameters
FICTURE uses numerous parameters at each step to ensure flexibility. NEDA simplifies data analysis by only requiring essential parameters in the input configuration file. Although some steps may require auxiliary parameters, NEDA adopts FICTURE's recommended defaults.
If you wish to customize these defaults, refer to the AUXILIARY PARAMS section in the step scripts and the FICTURE documentation, but proceed with caution due to potential risks.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42 | #=========================
# Mandatory Fields
#=========================
## Input files
input_transcripts=/path/to/the/transcripts/file ## Path to the input spatial digital gene expression (SGE) matrix in FICTURE-compatible TSV format.
input_features=/path/to/the/feature/file ## Path to the input feature file.
input_xyrange=/path/to/the/xyrange ## Path to the input meta file with minimum and maximum X Y coordinates.
## (Model-Specific) Input Hexagon-Indexed SGE matrix
# Those two analytical strategies in NEDA require input hexagon-indexed SGE matrix in different formats.
# Choose your analytical strategy first, then define its required hexagon-indexed SGE matrix.
input_hexagon_sge_ficture=/path/to/the/hexagon/indexed/sge/ficture ## (LDA-only) Path of hexagon-indexed SGE in the FICTURE-compatible TSV format.
input_hexagon_sge_10x_dir=/path/to/the/hexagon/indexed/sge/10x/dir ## (Seurat-only) Directory of hexagon-indexed SGE in the 10x genomics format, which should have features.tsv.gz, barcodes.tsv.gz, and matrix.mtx.gz.
## Output
output_dir=/path/to/the/output/directory/ ## Directory for output files: LDA results will be saved in ${output_dir}/LDA, and Seurat results wil be in ${output_dir}/Seurat."
prefix=<prefix_of_output_files> ## Prefix for output files. The output files will be named using both this prefix and the following parameters.
## Train model
train_model=<model_option> ## Define the analytical strategy. Options: "LDA", "Seurat".
## Params
major_axis=<X_or_Y> ## Typically, the major axis is the axis with a greater length. Options: "X", "Y". For instance, it is Y in the minimal testrun dataset whereas X in the shallow and deep liver datasets.
solo_feature=<solo_feature> ## Select the genome feature. Options: "gn": Gene; "gt": GeneFull. See details at https://github.com/alexdobin/STAR/blob/master/docs/STARsolo.md.
train_width=<train_width> ## The side length of the hexagon (in micrometers), e.g., 18.
fit_width=<projection_width> ## Projection width, suggest to use one the same as the train width, e.g., 18.
anchor_dist=<archor_distance> ## Anchor point distance (in micrometers), e.g., 4.
## (Model-Specific) params
## - LDA
nfactor=<number_of_factors> ## (LDA-only) Number of factors, e.g., 12. For 'Seurat+FICTURE' analysis, remove it when preparing the configuration file; nf will be defined after clustering.
train_n_epoch=<number_of_epoch> ## (LDA-only) Epochs for LDA training, e.g., 3. For "Seurat+FICTURE" analysis, use "NA" or remove it.
## - Seurat
#nFeature_RNA_cutoff=<the_optimal_cutoff> ## (Seurat-only) After evaluating the performance of different cutoffs, define the optimal cutoff aiming at removing noises.
res_of_interest=<the_optimal_resolution> ## (Seurat-only) After examining clustering results across all resolution settings, identify the optimal resolution.
#=========================
# Optional Fields
#=========================
#threads=<number_of_cpus> ## (Optional) A integer to indicate how many CPUs will be applied. If absent, 1 thread will be applied.
#seed=<an_integer> ## (Optional) A seed (integer, e.g., 2024030700) for reproducibility. This applies in the LDA factorization and choosing color maps. If omitted, a random seed will be utilized.
|