Skip to content

WORKLOG.rst

Latest commit

 

History

History
207 lines (161 loc) · 6.47 KB

WORKLOG.rst

File metadata and controls

207 lines (161 loc) · 6.47 KB

multiome-wf

2023-06-05

@Mira0507

  • workflow/WRAPPER_SLURM updated
    • to use snakemake_profile established by NIH HPC (https://github.com/NIH-HPC/snakemake_profile.git)
    • to remove lines using SNAKE_PID
  • workflow/Snakefile updated
    • directives resources added
    • res_use added to params of the rule cluster
      • enables users to set their clustering resolution in config.yaml file (see cluster in the yaml file)
    • workflow/cluster.Rmd updated accordingly
      • if statement added to choose clustering resolution between chooser-computed and user-provided values
  • workflow/config/multiome-config/config.yaml updated
    • dataset_size added to reveal whether input is a toy dataset
      • workflow/weighted_nn.Rmd updated accordingly
        • If input is toydataset, k.nn and k.range is adjusted in an if statement
        • This update is required to run FindMultiModalNeighbors() error-free with tiny input data
    • Indentation corrected
  • chooser updated
    • conda env updated
      • workflow/chooser/requirements.txt
      • workflow/chooser/env.yaml
    • multiple bugs fixed when handling toydataset in workflow/chooser_paral.Rmd and workflow/chooser_aggr.Rmd

2023-10-05

@Mira0507

  • Documentation updated
    • docs/configuration/config.rst updated
    • docs/configuration/config-yaml.rst updated

2024-07-02

@Mira0507

  • conda env tested

    macs2
graphviz
r-tidyverse
r-base
r-seurat
r-seuratobject
r-signac
r-matrix
snakemake<8
r-hdf5r
r-ggplot2
r-ggrepel
r-plotly
r-patchwork
r-devtools
r-remotes
bioconductor-bsgenome.hsapiens.ucsc.hg38
bioconductor-biovizbase
bioconductor-ensembldb
bioconductor-ensdb.hsapiens.v86
bioconductor-limma
bioconductor-ensdb.mmusculus.v79
bioconductor-genomicranges
samtools
r-data.table
bioconductor-rhdf5
python
pandas
rst2html5
alabaster

2024-07-03

@Mira0507

  • scripts updated
    • workflow/Snakefile: resources in diff_analysis rule
    • workflow/create_seurat.Rmd
      • code cleaned
      • cleaner printing in rendered files
    • workflow/qc.Rmd
      • cleaner printing in rendered files

2025-02-06

@Mira0507

  • Three rules added to workflow/Snakefile in order to create bigwig files using MACS2-called peaks

    • chromsizes: creates chromsizes
    • bigwig_signal: converts MACS2-created bdg files for signal to bigwig
    • bigwig_noise: converts MACS2-created bdg files for noise to bigwig

2025-02-12

@Mira0507

  • workflow/chooser_paral.Rmd updated to fix a bug

    • resolution set to 1.0 ended up deleting the results/chooser_run/unintegrated_3/res_1.0 directory after the completion of the following run

      rule chooser_paral:
    input: chooser_paral.Rmd, results/normalize_reduce_dims/unintegrated_3/unintegrated_3.rds, config/multiome-config/config.yaml
    output: results/chooser_run/unintegrated_3/res_1.0/report.html, results/chooser_run/unintegrated_3/res_1.0, results/chooser_run/unintegrated_3/silhouette_1.0.rds, results/chooser_run/unintegrated_3/frequency_grouped_1.0.rds, results/chooser_run/unintegrated_3/silhouette_grouped_1.0.rds, results/chooser_run/unintegrated_3/seurat_obj_1.0.rds
    jobid: 56
    reason: Missing output files: results/chooser_run/unintegrated_3/frequency_grouped_1.0.rds, results/chooser_run/unintegrated_3/silhouette_grouped_1.0.rds, results/chooser_run/unintegrated_3/silhouette_1.0.rds; Input files updated by another job: results/normalize_reduce_dims/unintegrated_3/unintegrated_3.rds
    wildcards: seurat=unintegrated_3, res=1.0
    threads: 12
    resources: tmpdir=/tmp, mem_mb=102400, mem_mib=97657, disk_mb=51200, disk_mib=48829, runtime=720

    • fix: set the resolution to 1 instead of 1.0 in config.yaml

      chooser:
  groups:
    unintegrated_1:
      npcs: 20
    unintegrated_2:
      npcs: 20
    unintegrated_3:
      npcs: 20
    integrated_1:
      npcs: 20
    integrated_2:
      npcs: 20
    integrated_3:
      npcs: 20
  resolutions:
    - 0.8
    - 1     # UPDATED!
    - 1.2
  silhouette:
    - silhouette
    - frequency_grouped
    - silhouette_grouped

2025-02-18

@Mira0507

  • workflow/chooser_aggr.Rmd updated to fix the following error:

    Error in bca.ci(boot.out, conf, index[1L], L = L, t = t.o, t0 = t0.o,  :
  estimated adjustment 'w' is infinite

    • bug: this error was raised when the boot::boot.ci is called with the following input vector: c(1, 0, 1, 0, 1, 0, 0). this error appeared to be raised by few unique input values (0 and 1), which results in a trouble with smoothing resampling distributions in bootstrapping.

      # workflow/chooser/R/pipeline.R
boot_median <- function(x, interval = 0.95, R = 25000, type = "bca") {
  # Define median to take data and indices for use with boot::
  med <- function(data, indices) {
    resample <- data[indices]
    return(median(resample))
  }

  # Calculate intervals
  boot_data <- boot::boot(data = x, statistic = med, R = R)
  boot_ci <- boot::boot.ci(boot_data, conf = interval, type = type)

  # Extract desired statistics
  ci <- list(
    low_med = boot_ci$bca[4],
    med = boot_ci$t0,
    high_med = boot_ci$bca[5]
  )
  return(ci)
}

    • fix: introduce small random noise to the input vector

      x_jittered <- x + rnorm(length(x), mean = 0, sd = 0.000001)
boot_median(x_jittered, type="bca")

2025-03-27

@Mira0507

  • Documentation updated