RGP

The repo contains the scripts to recreate the human analysis portion for the the manuscript: Regulatory activity is the default DNA state in eukaryotes. Using the raw data provided a user can create their own shuffled sequence datasets using Biasaway and then produce downstream plots. Detailed information about the set up can be found in methods section here: https://www.biorxiv.org/content/10.1101/2022.12.16.520785v1.full

Generating shuffled sequences

Exact datasets used in the paper can be found here: https://zenodo.org/records/10535706

To create shuffled sequence sets use test_hg38_seqs_196kb.fasta and biasaway (sample commands below). Test set regions (196kb each) with "N"s were filtered out to ensure proper shuffling and the surrounding flanking regions were added post shuffling.

Local Shuffling

To create local trinucleotide shuffled sequences use the following command with k = 3 (kmer), step size of 100bp and window size of 200bp:

biasaway w -f /project/st-cdeboer-1/iluthra/enformer_random_DNA/enformer_data/test_hg38_seqs_196kb.fasta -k 3 -s 100 -w 200 > /project/st-cdeboer-1/iluthra/enformer_random_DNA/enformer_data/test_hg38_seqs_196kb_trinuc_local.fasta

Global Shuffling

To create global trinucleotide shuffled sequences use the following command with k = 3 (kmer):

biasaway k -f /project/st-cdeboer-1/iluthra/enformer_random_DNA/enformer_data/test_hg38_seqs_196kb.fasta -k 3 > /project/st-cdeboer-1/iluthra/enformer_random_DNA/enformer_data/test_hg38_seqs_196kb_trinuc.fasta

Running Enformer to predict on shuffled sequence datasets

Enformer model weights and enviroment was installed using instructions from: https://github.com/google-deepmind/deepmind-research/tree/master/enformer

When predicting use the correctly padded test set sequence file: test_hg38_seqs_393kb_N_flank.fasta, random_sequences.fasta and other shuffled sequence datasets.

qsub enformer_multi.pbs

Running analysis pipeline to create the figures from the paper

Plots in the paper can be recreated individually using each script separately or all the analysis (except Fig 4b) can be made using RGP_plots.pbs script. Input and output paths will need to be changed.

qsub RGP_plots.pbs

To calculate the values used in Fig 4B. and Supplementary Figure 6 the script in the /Evolution_plots_12cellTypes/ folder will be used.

submitlog -m 10 -A st-cdeboer-1 -a [email protected] -o /scratch/st-cdeboer-1/iluthra/randomDNA/Analysis_outputs_06152023/H1_DNase_rep1_07052023/Evolution_plots_12cellTypes//doExample1.olog -t RGP_cell_types_annotation_file.csv ./submitlog.sh RGP_cell_types_annotation_file.csv

To create the barplots in Fig 4B. and Supplementary Figure 6 use: create_80percent_barplot.py

Questions

Feel free to contact Ishika Luthra ([email protected]) regardings questions or help.

References

Khan, A., Riudavets Puig, R., Boddie, P. & Mathelier, A. BiasAway: command-line and web server to generate nucleotide composition-matched DNA background sequences. Bioinformatics 37, 1607–1609 (2021).

Avsec, Ž. et al. Effective gene expression prediction from sequence by integrating long-range interactions. Nat Methods 18, 1196–1203 (2021).