{ "cells": [ { "cell_type": "markdown", "id": "70438228", "metadata": {}, "source": [ "# Data Preprocessing (Interactive Notebook)\n", "\n", "This notebook runs the repository's Step1→Step9 data-preprocessing steps interactively so you can execute, inspect and debug individual stages locally.\n", "\n", "Outputs from these steps will be written under `example/data_preprocessing` and downstream under `example/machine_learning` by default." ] }, { "cell_type": "code", "execution_count": null, "id": "2348a749", "metadata": {}, "outputs": [], "source": [ "%%bash\n", "git clone https://github.com/hlnicholls/GenePrioritiser\n", "cd GenePrioritiser\n", "\n", "conda env create -f GenePrioritiser_env.yml\n", "conda activate GenePrioritiser_env\n", "pip install --force-reinstall scikit-learn==1.4.2\n", "pip install --force-reinstall scipy==1.11.4\n", "pip install --force-reinstall numpy==1.23.0\n", "pip install scikit-optimize\n", "conda install -c bioconda bedtools htslib\n", "conda install -c conda-forge parallel\n", "pip install pybedtools intervaltree requests" ] }, { "cell_type": "code", "execution_count": 1, "id": "74317926", "metadata": {}, "outputs": [], "source": [ "import os\n", "os.chdir(\"/Users/hannahnicholls/GitHub/GenePrioritiser\")" ] }, { "cell_type": "code", "execution_count": 2, "id": "d9113571", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Using cached gene bed: /Users/hannahnicholls/GitHub/GenePrioritiser/utils/gencode_v19.genes.ext10000.sorted.bed\n", "Processing /Users/hannahnicholls/GitHub/GenePrioritiser/./example/data_preprocessing/input/Evangelou_30224653_DBP.txt.gz in /Users/hannahnicholls/GitHub/GenePrioritiser/tmp_annotation_Evangelou_30224653_DBP.txt_27680\n", "Parsing GWAS input and extracting SNP positions (Python)...\n", "Extracted 7160657 SNP positions from /Users/hannahnicholls/GitHub/GenePrioritiser/./example/data_preprocessing/input/Evangelou_30224653_DBP.txt.gz\n", "Running bedtools closest (this is the fast step, but can be I/O heavy on large GWAS)...\n", "Building gene map (line -> gene, distance)...\n", "Joining original rows with gene map and writing outputs via Python...\n", "Wrote annotated 7160657 SNPs to ./example/data_preprocessing/output/variants/Annotated_GWAS_DBP.csv\n", "Wrote variant data to ./example/data_preprocessing/output/variants/variant_data_DBP.csv\n", "Cleaning up temporary files: /Users/hannahnicholls/GitHub/GenePrioritiser/tmp_annotation_Evangelou_30224653_DBP.txt_27680\n", "Done for /Users/hannahnicholls/GitHub/GenePrioritiser/./example/data_preprocessing/input/Evangelou_30224653_DBP.txt.gz\n", "Processing /Users/hannahnicholls/GitHub/GenePrioritiser/./example/data_preprocessing/input/Evangelou_30224653_SBP.txt.gz in /Users/hannahnicholls/GitHub/GenePrioritiser/tmp_annotation_Evangelou_30224653_SBP.txt_27680\n", "Parsing GWAS input and extracting SNP positions (Python)...\n", "Extracted 7088121 SNP positions from /Users/hannahnicholls/GitHub/GenePrioritiser/./example/data_preprocessing/input/Evangelou_30224653_SBP.txt.gz\n", "Running bedtools closest (this is the fast step, but can be I/O heavy on large GWAS)...\n", "Building gene map (line -> gene, distance)...\n", "Joining original rows with gene map and writing outputs via Python...\n", "Wrote annotated 7088121 SNPs to ./example/data_preprocessing/output/variants/Annotated_GWAS_SBP.csv\n", "Wrote variant data to ./example/data_preprocessing/output/variants/variant_data_SBP.csv\n", "Cleaning up temporary files: /Users/hannahnicholls/GitHub/GenePrioritiser/tmp_annotation_Evangelou_30224653_SBP.txt_27680\n", "Done for /Users/hannahnicholls/GitHub/GenePrioritiser/./example/data_preprocessing/input/Evangelou_30224653_SBP.txt.gz\n", "Processing /Users/hannahnicholls/GitHub/GenePrioritiser/./example/data_preprocessing/input/Evangelou_30224653_PP.txt.gz in /Users/hannahnicholls/GitHub/GenePrioritiser/tmp_annotation_Evangelou_30224653_PP.txt_27680\n", "Parsing GWAS input and extracting SNP positions (Python)...\n", "Extracted 7088842 SNP positions from /Users/hannahnicholls/GitHub/GenePrioritiser/./example/data_preprocessing/input/Evangelou_30224653_PP.txt.gz\n", "Running bedtools closest (this is the fast step, but can be I/O heavy on large GWAS)...\n", "Building gene map (line -> gene, distance)...\n", "Joining original rows with gene map and writing outputs via Python...\n", "Wrote annotated 7088842 SNPs to ./example/data_preprocessing/output/variants/Annotated_GWAS_PP.csv\n", "Wrote variant data to ./example/data_preprocessing/output/variants/variant_data_PP.csv\n", "Cleaning up temporary files: /Users/hannahnicholls/GitHub/GenePrioritiser/tmp_annotation_Evangelou_30224653_PP.txt_27680\n", "Done for /Users/hannahnicholls/GitHub/GenePrioritiser/./example/data_preprocessing/input/Evangelou_30224653_PP.txt.gz\n" ] } ], "source": [ "!bash /Users/hannahnicholls/GitHub/GenePrioritiser/src/data_preprocessing/Step1_annotate_genes_bedtools.sh" ] }, { "cell_type": "code", "execution_count": 3, "id": "9ff8bfe6", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Repo root : /Users/hannahnicholls/GitHub/GenePrioritiser\n", "Variant directory : example/data_preprocessing/output/variants\n", "HGNC file : /Users/hannahnicholls/GitHub/GenePrioritiser/utils/hgnc_complete_set.txt\n", "Building HGNC synonym map...\n", "Loaded 102319 synonym mappings\n", "\n", "=== Harmonising Annotated_GWAS_* files ===\n", " Processing Annotated_GWAS_DBP.csv ... [OK, 428054 gene names changed]\n", " Processing Annotated_GWAS_PP.csv ... [OK, 423689 gene names changed]\n", " Processing Annotated_GWAS_SBP.csv ... [OK, 423690 gene names changed]\n", "\n", "=== Harmonising variant_data_* files ===\n", " Processing variant_data_DBP.csv ... [OK, 428054 gene names changed]\n", " Processing variant_data_PP.csv ... [OK, 423689 gene names changed]\n", " Processing variant_data_SBP.csv ... [OK, 423690 gene names changed]\n", "\n", "Done. All applicable files overwritten with HGNC-harmonised Gene symbols.\n" ] } ], "source": [ "!python /Users/hannahnicholls/GitHub/GenePrioritiser/src/data_preprocessing/Step2_harmonise_genes.py" ] }, { "cell_type": "code", "execution_count": 4, "id": "8a20f2dd", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "All files processed and merged. Output saved to: ./databases/variant_level/merged_gene_median_variant_measures.csv\n" ] } ], "source": [ "!python /Users/hannahnicholls/GitHub/GenePrioritiser/src/data_preprocessing/Step3_process_variant_level_data.py" ] }, { "cell_type": "code", "execution_count": 5, "id": "1ef583bf", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Number of unmapped entries: 0\n", "Sample unmapped entries:\n", "Empty DataFrame\n", "Columns: [protein1, protein2, coexpression, experiments, database, Gene1, Gene2]\n", "Index: []\n", "Interactors collected with max_depth=3: 19430 genes\n", "Direct only interactors (depth=1): 17554\n", "One-level interactors (depth=2): 19428\n", "Found 3 annotated GWAS file(s) to load:\n", " - ./example/data_preprocessing/output/variants/Annotated_GWAS_DBP.csv\n", " - ./example/data_preprocessing/output/variants/Annotated_GWAS_PP.csv\n", " - ./example/data_preprocessing/output/variants/Annotated_GWAS_SBP.csv\n", "Wrote intermediate least-likely genes to /Users/hannahnicholls/GitHub/GenePrioritiser/example/data_preprocessing/input/sampling/least_likely_intermediate.tsv (will not overwrite ./example/data_preprocessing/input/least_likely_genes.tsv)\n", "Applying additional filtering based on: ./example/data_preprocessing/input/BP_loci_Apr2020_LDr2-8_500kb.csv\n", "Filtered out 65 genes based on additional criteria (no genes with SNPs in any LD with any BP loci).\n", "Updated intermediate least-likely genes at /Users/hannahnicholls/GitHub/GenePrioritiser/example/data_preprocessing/input/sampling/least_likely_intermediate.tsv\n", "Least likely genes identified and saved.\n", "Number of least likely genes: 571\n", "Number of probable genes: 145\n", "Number of most likely genes: 51\n" ] } ], "source": [ "!python /Users/hannahnicholls/GitHub/GenePrioritiser/src/data_preprocessing/Step4_least_likely_gene_selection.py" ] }, { "cell_type": "code", "execution_count": 6, "id": "e42781c2", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Using intermediate least-likely genes file: /Users/hannahnicholls/GitHub/GenePrioritiser/example/data_preprocessing/input/sampling/least_likely_intermediate.tsv\n", "Identified 13758 genes with at least one P<0.01 SNP in EACH Annotated_GWAS file (per-file intersection)\n", "Probable genes: 254 -> 145 after applying per-file P<0.01 intersection filter\n" ] } ], "source": [ "!python /Users/hannahnicholls/GitHub/GenePrioritiser/src/data_preprocessing/Step6_identify_training_genes.py" ] }, { "cell_type": "code", "execution_count": 7, "id": "33902de9", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Found 5116 duplicate Gene entries in merged data - aggregating numeric features by mean to collapse duplicates.\n", "All merged databases saved to ./example/data_preprocessing/output/all_genes_merged_all_data.csv\n", "Training genes counts per label:\n", " least likely: 571\n", " probable: 145\n", " most likely: 51\n", "Joined data saved to ['./example/data_preprocessing/output/training_data_all_features.csv', './example/machine_learning/eda/input/training_data_all_features.csv']\n" ] } ], "source": [ "!python /Users/hannahnicholls/GitHub/GenePrioritiser/src/data_preprocessing/Step7_merge_all_databases_and_get_training_data.py" ] }, { "cell_type": "code", "execution_count": 8, "id": "f2f179e1", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Using intermediate least-likely genes file: /Users/hannahnicholls/GitHub/GenePrioritiser/example/data_preprocessing/input/sampling/least_likely_intermediate.tsv\n", "/Users/hannahnicholls/GitHub/GenePrioritiser/src/data_preprocessing/Step8_downsample_least_likely_genes.py:311: UserWarning: The palette list has more values (2) than needed (1), which may not be intended.\n", " sns.boxplot(data=[combined_pos['variant_count'], least_df['variant_count']], palette=['C0', 'C1'])\n", "Wrote diagnostic plots: /Users/hannahnicholls/GitHub/GenePrioritiser/example/data_preprocessing/input/sampling/variant_count_hist_seed42.png /Users/hannahnicholls/GitHub/GenePrioritiser/example/data_preprocessing/input/sampling/variant_count_box_seed42.png\n", "Wrote updated training_genes to ./example/data_preprocessing/input/training_genes.txt\n", "Wrote sampled least file: /Users/hannahnicholls/GitHub/GenePrioritiser/example/data_preprocessing/input/sampling/least_likely_sampled_seed42.tsv\n", "Wrote audit: /Users/hannahnicholls/GitHub/GenePrioritiser/example/data_preprocessing/input/sampling/least_likely_sampling_audit_seed42.json\n", "Sampled size: 51 target: 51\n", "Wrote training features to: ./example/data_preprocessing/output/training_data_all_features.csv\n", "Wrote EDA training input to: ./example/machine_learning/eda/input/training_data_all_features.csv\n" ] } ], "source": [ "!python /Users/hannahnicholls/GitHub/GenePrioritiser/src/data_preprocessing/Step8_downsample_least_likely_genes.py" ] }, { "cell_type": "code", "execution_count": 9, "id": "26b264fa", "metadata": {}, "outputs": [], "source": [ "!python /Users/hannahnicholls/GitHub/GenePrioritiser/src/data_preprocessing/Step9_subset_genes_to_prioritise.py" ] }, { "cell_type": "code", "execution_count": null, "id": "58784add", "metadata": {}, "outputs": [], "source": [] } ], "metadata": { "kernelspec": { "display_name": "GenePrioritiser_env", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.9.19" } }, "nbformat": 4, "nbformat_minor": 5 }