Skip to content

Latest commit

 

History

History
69 lines (41 loc) · 6.51 KB

workflow.md

File metadata and controls

69 lines (41 loc) · 6.51 KB

Contributors

Jake Lee

Lucy Jung

Table of contents

The computational scripts described here are available in the R folder.

Creating metadata

Here we share the scripts for creating the R dataframes summarizing the clinical and genomic information. The three output files (List.patients_summary.final.txt, List.purple.final.txt, and List.hrd_status.final.txt) are available in the Data folder.

Identifying focal amplifications

To identify focally amplified regions and to associate structural variations (SVs) to the boundaries of amplicons, we used HMF_definition_amp_segment.R script. This function requires the segmented allelic copy number and structural variation information produced by the HMF bioinformatic pipeline as the input as well as the dataframe summarizing the clinical information (the output from the previous section).

The output files from the HMF_definition_amp_segment.R were further analyzed by the scripts available in HMF_collecting_amp_segment_descriptive_analyses.R. Through this step, the SVs at the amplicon boundaries were summarized and annotated. Then, the copy number of the adjacent segments were analyzed to select the amplicons bordered by the unamplified segments for downsteam analyses.

Same analysis for the PCAWG dataset was performed using the PCAWG_definition_amp_segment.R and PCAWG_collecting_amp_segment_descriptive_analyses. Figure 5 was generated based on this analysis.

Epigenomic association

The following functions can be loaded by running association.with.epigenomics.data.R. The relavant data files are found in the Data folder. The nessary input files for the functions are typically set as default. Figure 3 was generated based on this analysis.

To determine which epigenomic features were associated with the early SV events initiating the focal amplifications, we modeled the location of the amplicon boundaries/SVs with various epigenomic features. The function association.with.chromatin.features takes the coordinates of factors, boundary positions, and numbers of bindings and boundaries as inputs and computes the enrichment p-values by the Lasso regression.

The function comparison.er.e2.controlcompares the distributions of ERa binding intensity in E2-treated MCF7 cells and non-treated MCF7 cells and annotates major amplicon boundary hotspots.

To study the association between the recurrence of SV breakpoints and ERa binding in E2-treated cells in unamplified regions, we calculated the recurrence of patients harboring SV breakpoints for each bin and the ERa binding in the E2-treated MCF7 cells. The function association.recurrence.e2.er.non.amp takes the information of the recurrencee, the ERa binding, and displays the increase of the percentage of regions with ERa binding with higher recurrences.

To analyze the association between breakpoints by ER treatmented HTGTS experiments (see below) and epigenomic features, we modeled the location of the breakpoints with various epigenomic features. The function association.with.chromatin.features.htgts takes the coordinates of factors, breakpoints, and numbers of bindings and breakpoints and computes the enrichment p-values by the Lasso regression.

HTGTS

The raw data from the HTGTS experiments is available at GEO GSE227369. The raw data was primarily analyzed using the scripts in the HTGTS_data_processing.R. The main output file from this script, HTGTS.lograt.average.allcells.twotarget.gene.v2.corrected.allinfo.txt, is available in the Data folder.

The scripts used for downstream analyses of the HTGTS dataset are available in the HTGTS_annotation_visualization.R. The raw output from the GSEA analysis, GSEA.report_for_na_pos_1650575918912.tsv, is available in the Data. Relevant for Figure 3.

Timing analysis

To analyze the timing of copy-number gains in the breast cancer genomes, we used two different methods, relative and absolute timing of the segments as described in the manuscript. First, relative timing was analyzed based on the methods used in the PCAWG analysis (MutationTimeR package was used) using our custom code TimeR_CNA_WGD.R. Synchronicity of the copy-number gains were assessed using TimeR_synchronicity_analysis.R.

For absolute timing, we determined the burden of pre-amplification mutation for each amplified genomic segment using VEClonal_calculation_binomial.R. Using the output of this analysis, we estimated the timing of major copy-number gains as well as the non-bridge arm gains from the select cases with TB amplification based on the scripts collected in VEClonal_timing_analysis.R. Figure 4 was generated based on this analysis.

Transcriptome analysis

We devised a score reflecting mRNA expression of the estrogen-responsive genes and applied this to the 263 breast cancer cases with available RNA sequencing datasets. The scripts used in this analysis are available in ERalpha_transcriptome_activity.R. Relevant for Figure 4.

We also analyzed the expression of amplified genes and correlated with their knockout phenotype in the CRISPR screen data. Related scripts are available in Amplified_genes_expression.R.

Data visualization

To illustrate structural variations and their associated copy number information, We used SVsketch_20chrom_HMF_clean.R. This was used in Figures 1, 2, and 4.

To visualize the genomic rearrangement landscape in 780 breast cancers, we used Plot_SV_matrix.R script. Relevant for Figure 1.

To visualize the amplified regions, their boundaries, and the association with the HTGTS breakpoints, we used Plot_amplicon_HTGTS.R script.

We used Oncoprint_clinical_signatures.R script to illustrate the genomic alteration landscape (Extended Data Fig. 2d). The output files were integrated using the Adobe Illustrator.