Skat

Dockerfile

@@ -61,6 +61,15 @@ RUN git clone https://github.com/froelich/sf-relate . && \
     rm -rf .git
 
 
+# -------------------- sf-skat -------------------- #
+FROM go AS sf-skat
+
+RUN git clone https://github.com/swanhong/secure-skat . && \
+    git checkout b79ed20 && \
+    go build -o secure-rvas secure-rvas.go && \
+    rm -rf .git
+
+
 # -------------------- sfkit-proxy -------------------- #
 FROM go AS sfkit-proxy
 
@@ -201,7 +210,7 @@ RUN microdnf install -y proxychains-ng && \
 
 ENV HOME=${WORK} \
     OPENSSL_FORCE_FIPS_MODE=1 \
-    PATH="${WORK}/.venv/bin:$PATH:${WORK}:${WORK}/sfgwas:${WORK}/sf-relate:${WORK}/sfgwas-lmm/scripts" \
+    PATH="${WORK}/.venv/bin:$PATH:${WORK}:${WORK}/sfgwas:${WORK}/sf-relate:${WORK}/sfgwas-lmm/scripts:${WORK}/sf-skat" \
     PYTHONPATH="${WORK}/.venv/lib/python3.12/site-packages:${WORK}/.venv/lib64/python3.12/site-packages" \
     PYTHONUNBUFFERED=TRUE \
     PYTHONWARNINGS="ignore:pkg_resources is deprecated as an API:UserWarning" \
@@ -215,6 +224,7 @@ COPY --from=secure-gwas --chown=$USER ${WORK}           ./secure-gwas/
 COPY --from=sfgwas      --chown=$USER ${WORK}           ./sfgwas/
 COPY --from=sfgwas-lmm  --chown=$USER ${WORK}           ./sfgwas-lmm/
 COPY --from=sf-relate   --chown=$USER ${WORK}           ./sf-relate/
+COPY --from=sf-skat     --chown=$USER ${WORK}/*-rvas    ./sf-skat/
 COPY --from=sfkit-proxy --chown=$USER ${WORK}/*-proxy   ./
 
 COPY --from=sfkit --chown=$USER ${WORK}/dist/sfkit*.whl ./

install.sh

@@ -62,14 +62,15 @@ rm -rf ~/.local/sf-relate && mv sf-relate ~/.local/
 rm -rf ~/.local/secure-dti && mv secure-dti ~/.local/
 rm -rf ~/.local/secure-gwas && mv secure-gwas ~/.local/
 rm -rf ~/.local/sfgwas-lmm && mv sfgwas-lmm ~/.local/
+rm -rf ~/.local/sf-skat && mv sf-skat ~/.local/
 echo
 
 # check if ldd version is at least 2.34
 glibc_minor_ver=$(ldd --version | awk 'NR==1{print $NF}' | cut -d. -f2)
 if [ "${glibc_minor_ver}" -lt 34 ]; then
   echo Patching sfkit binaries...
   mkdir -p ~/.local/lib/ && mv lib/* ~/.local/lib/
-  for p in ~/.local/bin/sfkit-proxy ~/.local/sfgwas/sfgwas ~/.local/sf-relate/sf-relate ~/.local/secure-dti/mpc/code/bin/* ~/.local/secure-gwas/code/bin/* ; do
+  for p in ~/.local/bin/sfkit-proxy ~/.local/sfgwas/sfgwas ~/.local/sf-relate/sf-relate ~/.local/sf-skat/secure-rvas ~/.local/secure-dti/mpc/code/bin/* ~/.local/secure-gwas/code/bin/* ; do
     patchelf --set-interpreter ~/.local/lib/ld-linux-x86-64.so.2 "$p"
   done
   echo
@@ -90,6 +91,7 @@ $PWD/.venv/bin:\
 \$HOME/.local/sfgwas:\
 \$HOME/.local/sfgwas-lmm:\
 \$HOME/.local/sf-relate:\
+\$HOME/.local/sf-skat:\
 \$HOME/.local/secure-dti/mpc/code/bin:\
 \$HOME/.local/secure-gwas/code/bin\
 \"" >> "$rc"

sfkit/protocol/register_data.py

@@ -58,6 +58,8 @@ def register_data(
             data_path = validate_sfrelate(doc_ref_dict, username, data_path, role)
         elif study_type == "Secure-DTI":
             data_path = validate_dti(doc_ref_dict, username, data_path, role)
+        elif study_type == "SF-SKAT":
+            data_path = validate_skat(doc_ref_dict, username, data_path, role)
         else:
             raise ValueError(f"Unknown study type: {study_type}")
 
@@ -288,6 +290,61 @@ def validate_dti(doc_ref_dict: dict, username: str, data_path: str, role: str) -
     return data_path
 
 
+def validate_skat(
+    doc_ref_dict: dict, username: str, data_path: str, role: str
+) -> str:
+    """
+    Validate data for the SF-SKAT workflow.
+
+    SF-SKAT does not run data preparation itself: data_path must already contain
+    a `prepared/` directory tree produced by running `secure-rvas prepare`
+    out-of-band (see https://github.com/swanhong/secure-skat). data_path is the
+    directory that *contains* `prepared/` (i.e. secure-rvas's `run_dir`).
+
+    Only called for roles 1 (Cohort A) and 2 (Cohort B) -- role 0 (auxiliary)
+    holds no data and never reaches this function.
+    """
+    data_path = validate_data_path(data_path)
+
+    if data_path == "demo" or (constants.IS_DOCKER and doc_ref_dict["demo"]):
+        using_demo()
+
+    ancestries = [
+        a.strip()
+        for a in doc_ref_dict["parameters"]["ancestries"]["value"].split(",")
+        if a.strip()
+    ]
+    chromosomes = [
+        c.strip()
+        for c in doc_ref_dict["parameters"]["chromosomes"]["value"].split(",")
+        if c.strip()
+    ]
+    cohort = "A" if role == "1" else "B"
+
+    for ancestry in ancestries:
+        for chromosome in chromosomes:
+            chr_dir = os.path.join(data_path, "prepared", ancestry, f"chr{chromosome}")
+            condition_or_fail(
+                os.path.isfile(os.path.join(chr_dir, "genes.txt")),
+                f"Could not find {chr_dir}/genes.txt",
+            )
+            condition_or_fail(
+                os.path.isfile(os.path.join(chr_dir, "block_sizes.txt")),
+                f"Could not find {chr_dir}/block_sizes.txt",
+            )
+            cohort_dir = os.path.join(chr_dir, cohort)
+            condition_or_fail(
+                os.path.isfile(os.path.join(cohort_dir, "cov.txt")),
+                f"Could not find {cohort_dir}/cov.txt",
+            )
+            condition_or_fail(
+                os.path.isfile(os.path.join(cohort_dir, "pheno.txt")),
+                f"Could not find {cohort_dir}/pheno.txt",
+            )
+
+    return data_path
+
+
 def validate_geno_binary_file_prefix(geno_binary_file_prefix: str) -> str:
     if not geno_binary_file_prefix:
         if constants.IS_DOCKER and os.path.exists("/data/geno"):

sfkit/protocol/run_protocol.py

@@ -9,6 +9,7 @@
 from sfkit.utils.sfgwas_lmm_protocol import run_sfgwas_lmm_protocol
 from sfkit.utils.sfgwas_protocol import run_sfgwas_protocol
 from sfkit.utils.sfrelate_protocol import run_sfrelate_protocol
+from sfkit.utils.sfskat_protocol import run_sfskat_protocol
 
 
 def run_protocol(
@@ -93,6 +94,8 @@ def run_protocol(
             run_sfrelate_protocol(role, demo)
         elif study_type == "Secure-DTI":
             run_dti_protocol(role, demo)
+        elif study_type == "SF-SKAT":
+            run_sfskat_protocol(role, demo)
         else:
             raise ValueError(f"Unknown study type: {study_type}")
     else:

sfkit/utils/constants.py

@@ -39,7 +39,7 @@ def is_installed(binary: str) -> bool:
     if os.environ.get("PATH"):
         os.environ[
             "PATH"
-        ] += f":{EXECUTABLES_PREFIX}bin:{EXECUTABLES_PREFIX}sfgwas:{EXECUTABLES_PREFIX}sf-relate:{EXECUTABLES_PREFIX}secure-gwas/code/bin"
+        ] += f":{EXECUTABLES_PREFIX}bin:{EXECUTABLES_PREFIX}sfgwas:{EXECUTABLES_PREFIX}sf-relate:{EXECUTABLES_PREFIX}sf-skat:{EXECUTABLES_PREFIX}secure-gwas/code/bin"
 
 SFKIT_PREFIX = "sfkit: "
 OUT_FOLDER = os.path.join(os.environ.get("SFKIT_DIR", ""), "out")