Skip to content

Homework 1

Homework 1: Fundamentals of scRNA-seq analysis in Python

Deadline: Friday, Jan 30th 2026 11:59pm

Background

We discussed hematopoiesis and the types of cells found in peripheral blood mononuclear cells (PBMCs) in Lecture 3. We also walked through the analysis of a human PBMC scRNA-seq data set, from CellRanger output to a cleaned anndata object decorated with cell type annotations for each cell. In this homework, you are going to explore what scRNA-seq tells us about PBMC cell type composition, and you are going to explore the impact of varying steps in the analysis pipeline on your estimate of cell type composition.

Cell type composition

You should think about and compute Cell type composition relative to the total number of cells sampled. For example, after you perform QC on Sample 1, maybe your anndata object has 10,000 cells. Downstream analysis (i.e. all the steps from filtering to annotation) results in 7500 cells labeled as 'T cell'. In this case, the T cell fraction of the cell composition is 75%.

Cell types in PBMCs and their distinctive genes

Composition, as claimed in literature
  • T cell: 70%
  • B cell: 15%
  • Natural killer (NK) cells: 10%
  • Monocyte: 5%
  • Dendritic cells: 1%
  • Yes, the total is 101%. Don't look at me. I am just reporting what is listed in the literature.
Marker genes
Monocyte
  • LYZ (Lysozyme), CD14, CD68
  • sub-types:
    • Classical: CD14, LYZ, S100A8/S100A9, CCR2
    • Intermediate: FCGR3A (CD16), CD163, IL1B
    • non-classical: CX3CR1, FCGR3A (CD16), CCR5
Natural killer cell
  • NCAM1 , KIR2DL1, KIR2DL3, KIR2DL4, KIR3DL1, KIR3DL2, NKG2A/C/E (KLRC1/KLRC2/KLRC3), NKG2D (KLRK1), GNLY, GZMB
  • Note that some are also expressed by subsets of T cell
B cell
  • CD19, CD79A, CD79B, CD20
  • Will leave sub-type exploration to student
T cell
  • CD3D, CD3E, CD3G
  • Will leave sub-type exploration to student
Dendritic cell
  • FLT3, CD11C CD1C
  • CD123 & CLEC4C (plasmacytoid dendritic cell)
Granulocytes and Megakaryocytes
  • Not widely recognized as occurring in PBMC
  • If you suspect that these cells are present in the data, then you should identify and include makers thereof in your analysis

Data

This is the same Sample 1 as we used in Lecture 4. Here are some more details:

  • 11,769 PBMCs from a healthy Donor. Data generated using 10X genomics v3 chemistry and processed with Cell Ranger 3.0.0
  • Sequenced on Illumina NovaSeq with approximately 54,000 reads per cell
  • 10X Genomics web page for this data
  • URL to h5 file:
    • https://cf.10xgenomics.com/samples/cell-exp/3.0.0/pbmc_10k_v3/pbmc_10k_v3_filtered_feature_bc_matrix.h5

Your mission

Analyze the provided data to address the questions listed below

  1. What cell types are present in the data and in what proportions?

    Deliverables:

    • new .obs column indicating cell type
    • appropriate visualization (i.e. UMAP with clusters labeled, supported by dotplot that shows expression of marker genes)

  2. How does this estimate of cell composition compare to prior literature? To evaluate this rapidly and efficiently, generate a figure that depicts cell type composition as a stacked barplot. Your figure should show one stacked bar representing the scRNAseq cell type proportions, and one that represents the proportions from the literature as listed above. To be clear, each rectangle in stacked bar should reflect the percent of a sample that is annotated as cell type X.

    Deliverables:

    • Show your code
    • The figure

  3. What are the transcriptional signatures of each of the PBMC cell types? We know about some genes that are preferentially expressed in each PBMC cell type (in fact, you used some of these to perform cluster annotation). But one of the benefits of performing genome-wide analyses is that it can identify genes previously unlinked to the phenomenon under study. Your task here is to identify all genes that are preferentially expressed in each PBMC cell type.

    Deliverables:

    • Show your code
    • The lists of signature genes for each cell type.
    • Visualization: plot the cell type signatures using the builtin scanpy functions for heatmaps, matrixplot and dotplot. Use your judgement to select n, the top n set of genes to display for each cell type. Indicate in each figure the marker genes listed above.

ALL SUBMISSIONS MUST...

... include all code necessary to replicate the results, and must adhere to posted guidelines for code documentation, cogent written reports, and meaningful figures.