Add new method: LINGER

config/config.yaml

@@ -52,6 +52,10 @@ methods:
     inferelator:
         modular: False
         ext: 500000
+    linger:
+        modular: False
+        version: 'LINGER'
+        mode: 'parallel'
     pando:
         modular: True
         exclude_exons: 'TRUE'
@@ -108,6 +112,7 @@ colors:
         granie: '#97de8a'
         hummus: '#d52727'
         inferelator: '#ff9796'
+        linger: '#17a589'
         pando: '#9467bc'
         scdori: '#c4afd4'
         scenicplus: '#8c554a'
@@ -194,6 +199,7 @@ method_names:
   grnboost: GRNBoost2
   hummus: HuMMuS
   inferelator: Inferelator3.0
+  linger: LINGER
   pando: Pando
   pearson: Pearson
   random: Random

workflow/Snakefile

@@ -97,6 +97,7 @@ include: 'rules/dbs/gst.smk'
 include: 'rules/dbs/c2g.smk'
 include: 'rules/dbs/prt.smk'
 include: 'rules/dbs/sim.smk'
+include: 'rules/dbs/linger_prior.smk'
 
 # Datasets
 include: 'rules/dts/pbmc10k.smk'
@@ -133,6 +134,7 @@ include: 'rules/mth/scdori.smk'
 include: 'rules/mth/scenic.smk'
 include: 'rules/mth/scgpt.smk'
 include: 'rules/mth/scmtni.smk'
+include: 'rules/mth/linger.smk'
 
 # Analyses
 include: 'rules/anl/metrics/prior.smk'

workflow/envs/linger.def

@@ -0,0 +1,90 @@
+Bootstrap: docker
+From: ubuntu:20.04
+
+
+%environment
+    export PATH=/opt/:$PATH
+    . "/opt/conda/etc/profile.d/conda.sh"
+    . "/opt/conda/etc/profile.d/mamba.sh"
+    conda activate env
+
+%post
+
+    export DEBIAN_FRONTEND=noninteractive
+    apt update -y
+
+    # basic packages
+    apt-get install -y --no-install-recommends tzdata
+    apt install -y build-essential  \
+                   gcc \
+                   g++ \
+                   libstdc++6 \
+                   cmake \
+                   wget \
+                   curl \
+                   git \
+                   zlib1g-dev \
+                   libbz2-dev \
+                   liblzma-dev \
+                   libcurl4-openssl-dev \
+                   libssl-dev \
+                   libxml2-dev \
+                   libcairo2-dev \
+                   libxt-dev \
+                   libopenblas-dev \
+                   ca-certificates \
+                   bedtools
+
+    # conda
+    wget -O Miniforge3.sh "https://github.com/conda-forge/miniforge/releases/latest/download/Miniforge3-$(uname)-$(uname -m).sh"
+    bash Miniforge3.sh -b -p "/opt/conda"
+    rm Miniforge3.sh
+    . "/opt/conda/etc/profile.d/conda.sh"
+    . "/opt/conda/etc/profile.d/mamba.sh"
+    conda activate
+
+    # Create env
+    mamba create -y -n env -c pytorch -c bioconda -c conda-forge \
+    python=3.10 \
+    pip \
+    setuptools \
+    wheel \
+    cython \
+    pytorch=2.5.1 \
+    torchvision \
+    torchaudio \
+    cpuonly \
+    pybedtools==0.10.0 \
+    bedtools \
+    homer \
+    r-base==4.3 \
+    r-ggplot2 \
+    r-tidyr \
+    r-tidyverse \
+    rpy2 \
+    jupyterlab \
+    notebook
+
+    # Install LingerGRN and dependencies
+    conda activate env
+    pip install \
+        scipy==1.11.3 \
+        numpy==1.24.3 \
+        pandas==2.0.3 \
+        shap==0.42.0 \
+        scikit-learn==1.3.0 \
+        joblib==1.3.2 \
+        matplotlib==3.8.0 \
+        seaborn==0.13.0 \
+        statsmodels==0.14.1 \
+        umap-learn \
+        scanpy==1.9.5 \
+        anndata==0.9.2 \
+        "mudata<0.2.4" \
+        LingerGRN==1.106
+
+    # Remove cache for lighter containers
+    pip cache purge
+    conda clean -a -y
+    apt-get clean
+    rm -rf /var/lib/apt/lists/*

workflow/rules/dbs/linger_prior.smk

@@ -0,0 +1,20 @@
+import os
+
+rule linger_prior:
+    threads: 1
+    singularity: 'workflow/envs/linger.sif'
+    input: 'workflow/envs/linger.sif'
+    output:
+        archive='dbs/lingerGRN/data_bulk.tar.gz',
+        dir=directory('dbs/lingerGRN/data_bulk')
+    params:
+        dir_name=lambda wildcards, output: os.path.dirname(output.archive)
+    shell:
+        """
+        mkdir -p {params.dir_name}
+        wget -nv --load-cookies /tmp/cookies.txt \
+        "https://drive.usercontent.google.com/download?export=download&confirm=$(wget --quiet --save-cookies /tmp/cookies.txt --keep-session-cookies --no-check-certificate 'https://drive.usercontent.google.com/download?id=1jwRgRHPJrKABOk7wImKONTtUupV7yJ9b'  -O- | sed -rn 's/.*confirm=([0-9A-Za-z_]+).*/\1/p')&id=1jwRgRHPJrKABOk7wImKONTtUupV7yJ9b" \
+        -O '{output.archive}'
+        rm -f /tmp/cookies.txt
+        tar -xzf '{output.archive}' -C '{params.dir_name}'
+        """

workflow/rules/dbs/tss.smk

@@ -1,4 +1,4 @@
-localrules: gen_tss_celloracle, gen_tss_crema, gen_tss_dictys, gen_tss_figr, gen_tss_granie, gen_tss_pando, gen_tss_scenicplus, gen_tss_scdori, gen_tss_scmtni, gen_tss_promoters
+localrules: gen_tss_celloracle, gen_tss_crema, gen_tss_dictys, gen_tss_figr, gen_tss_granie, gen_tss_linger, gen_tss_pando, gen_tss_scenicplus, gen_tss_scdori, gen_tss_scmtni, gen_tss_promoters
 
 
 rule gen_tss_celloracle:
@@ -83,6 +83,21 @@ rule gen_tss_inferelator:
         cp {input} {output}
         """
 
+rule gen_tss_linger:
+    threads: 1
+    singularity: 'workflow/envs/gretabench.sif'
+    input:
+        gtf=rules.gen_genome_inferelator.output.gtf,
+        genes='dbs/lingerGRN/data_bulk/bulk_gene_all.txt'
+    output: 'dbs/hg38/gen/tss/linger.bed.gz'
+    shell:
+        """
+        python workflow/scripts/dbs/gen/tss/linger.py \
+        -i {input.gtf} \
+        -g {input.genes} \
+        -o {output}
+        """
+
 rule gen_tss_pando:
     threads: 1
     singularity: 'workflow/envs/pando.sif'

workflow/rules/mth/linger.smk

@@ -0,0 +1,39 @@
+rule mdl_o_linger:
+    threads: 24 
+    singularity: 'workflow/envs/linger.sif'
+    input:
+        img='workflow/envs/linger.sif',
+        linger_GRN=rules.linger_prior.output.dir,
+        mdata=rules.extract_case.output.mdata
+    output:
+        dir=directory('dts/{org}/{dat}/cases/{case}/runs/linger/'),
+        out='dts/{org}/{dat}/cases/{case}/runs/o_linger.o_linger.o_linger.o_linger.mdl.csv'
+    params:
+        version=config['methods']['linger']['version'],
+        mode=config['methods']['linger']['mode'],
+        organism=lambda w: config['dts'][w.dat]['organism'],
+        script='workflow/scripts/mth/linger/linger.sh'
+    resources:
+        mem_mb=lambda wildcards, attempt: restart_mem(wildcards, attempt) * 2,
+        runtime=(
+            30 if config['methods']['linger']['version'] == 'baseline'
+            else 140 if config['methods']['linger']['mode'] == 'parallel'
+            else 360
+        )
+    shell:
+        """
+        mkdir -p {output.dir}
+        set -e
+        timeout $(({resources.runtime}-20))m \
+        bash {params.script} \
+        --linger_GRN {input.linger_GRN} \
+        --out_dir {output.dir} \
+        --path_mdata {input.mdata} \
+        --version {params.version} \
+        --genome {params.organism} \
+        --mode {params.mode} \
+        --path_out {output.out}
+        if [ $? -eq 124 ]; then
+            awk 'BEGIN {{ print "source,target,score,pval" }}' > {output.out}
+        fi
+        """

workflow/scripts/dbs/gen/tss/linger.py

@@ -0,0 +1,42 @@
+import argparse
+import gzip
+import re
+import pandas as pd
+
+parser = argparse.ArgumentParser()
+parser.add_argument('-g', '--path_genes', required=True)
+parser.add_argument('-i', '--path_gtf', required=True)
+parser.add_argument('-o', '--path_out', required=True)
+args = parser.parse_args()
+
+gene_set = set(pd.read_csv(args.path_genes, header=None)[0].tolist())
+print(f"LINGER gene universe: {len(gene_set)} genes")
+
+records = []
+with gzip.open(args.path_gtf, 'rt') as f:
+    for line in f:
+        fields = line.rstrip('\n').split('\t')
+        if fields[2] != 'transcript':
+            continue
+        gene_name = re.search(r'gene_name "([^"]+)"', fields[8]).group(1)
+        records.append([fields[0], int(fields[3]), gene_name])
+
+# mirror `get_TSS_ensembl` in LINGER_tr_fast.py 
+df = pd.DataFrame(records, columns=['chr', 'start', 'symbol'])
+df = df.groupby(['chr', 'symbol'], as_index=False)['start'].min()
+df = df[df['symbol'] != '']       
+df['start'] -= 1                   
+
+std_chroms = [f'chr{i}' for i in range(1, 23)] + ['chrX']
+df = df[df['symbol'].isin(gene_set) & df['chr'].isin(std_chroms)]
+print(f"Matched {df['symbol'].nunique()}/{len(gene_set)} genes")
+
+bed = pd.DataFrame({
+    'chr':   df['chr'],
+    'start': df['start'],
+    'end':   df['start'],   
+    'name':  df['symbol']
+})
+bed = bed.sort_values(['chr', 'start'])
+bed.to_csv(args.path_out, sep='\t', index=False, header=False, compression='gzip')
+print(f"Written {len(bed)} records to {args.path_out}")