This repository provides analysis and processed data for our paper: An NFIX mediated regulatory network governs the balance of hematopoietic stem and progenitor cells during hematopoiesis

Seurat obj is manual_label.RData, can be found in the NCBI GEO database: GSE166922.

Figure 1 shows results for cell clustering and cell type annotation. Protein surface markers and known marker gene expression (panel C, D) were used to define cell types in panel B. Code see scRNA_seq/cite_seq_analysis.ipynb and scRNA_seq/3_7_marker_genes.ipynb

Figure 2 shows differential abundance analysis. Code for panel A, B: scRNA_seq/Figure2.ipynb. Code for panel C scRNA_seq/GSEApy.ipynb. GSEA results are scRNA_seq/GSEA*csv

During revision, we also calculated the following (see Code scRNA_seq/differential_abundance.ipynb):

1. KDE for different bandwidth

2. DAseq analysis for different resolutions

Initial Data Quality are provided in HPC5_data_analysis/**/*html and HPC7_data_analysis/**/*html

NFIX IDR peaks: HPC5_data_analysis/ChIP_seq/NFIX_IDR.noko.st.bed

PU.1 IDR peaks: HPC5_data_analysis/ChIP_seq/PU1_IDR.st.bed

ChIP-seq signal plots (e.g., Fig3.A) were generated using fold enrichment signal from MACS2 and NFIX IDR peaks. (Data can be found in the GEO database)

Pie chart showing peak distribution (e.g., Fig3.B) was generated using homer annotatePeaks

Motif enrichment (e.g., Fig3.C) was generated using homer findMotifsGenome.pl

Fig3.D and Fig3.E data (except NFIX) were from HPC7, see HPC7_data_analysis/ChIP_seq/. SRA ID is provided in HPC7_data_analysis/ChIP_seq/sra*list. We reprocessed these ChIP-seq data using our pipeline, if you need these bw files, just let us know.

HPC5 footprint signal (e.g., Fig4.B) was generated using HINT ATAC.

HPC5 super-enhancer in silico prediction code is here: HPC5_data_analysis/super_enhancer/SE.lsf, the resulting bed file is HPC5_data_analysis/super_enhancer/merged_SE.bed

NFIX-PU1 footprint and motif distance analysis: HPC5_data_analysis/NFIX_PU1_motif_footprint_analysis/ecdf-with-random.ipynb

Pipeline shown in Fig5 contains the following steps (in folder integrative_analysis).

1. Gathering ATAC-seq data from different blood lineages: step1_S3norm_ATAC

2. Together with NFIX and PU.1 motif, to predict new NFIX-PU.1 co-binding sites in other blood lineages: step2_Catchitt_peak_prediction

3. Direct target gene finding based on scRNA-seq gene expression (Fig5.B): step3_direct_targets

• captureC data: target.5kb.bed (extended +/- 5kb) is from Table S5 in https://www.sciencedirect.com/science/article/pii/S2211124719315566?via%3Dihub

• TAD data: GSE119347_BMHSC_TADs.bed.gz

• scRNA-seq DEG: *DEG.orig.ident.logFC02.FDR001.tsv

4. GO enrichment for direct target genes (Fig5.C): step4_GO_enrichment