search

califano-lab / single-cell-pipeline

40 stars 8 forks source link

HCA data testing #5

Open somnathtagore opened 6 years ago

somnathtagore commented 6 years ago

Census of Immune Cells: Profiling of immunocytes by single cell RNA-seq for understanding human health and disease

Species: Homo sapiens Organ: Umbilical cord blood and bone marrow Method: 10x Cell count: ~530,000 cells

Download data: Raw Counts Matrix - Cord Blood, Raw Counts Matrix - Bone Marrow Data Format: .hd5 (created by CellRanger)

Data Preparation: Convert .hd5 into .rda file, with only one matrix object.

Need to do: Check out the CellRanger packages for handling .hd5 files.

somnathtagore commented 6 years ago

Checked CellRanger packages.

Converted the raw counts matrix to .rda with one matrix object. Issues:

  1. Error in matrix handling by ExpressionPreparation.R
  2. Need to create a filtered-gene-matrix.h5 file
  3. CellRanger considers GRCh38 genome
  4. Can't allocate memory.
somnathtagore commented 6 years ago

HCA, blood data, CZI…. Contd.

  1. Possibly Naive data as they ask to filter out low quality cells before analysis. As such, nothing is mentioned in the metadata sheet,
  2. Ignored the low quality filtration step for now.
  3. Considered only 1:10,1:10 for genes x cells due to memory allocation.
  4. Converted Ensembl IDs to Gene IDs -> None of the gene ids match -> Possibly, version issue.