Merge pull request #391 from rachelse/gpu_sooyoung

initial GPU support for monomer search
steineggerlab · Jan 5, 2025 · fb2b4ac · fb2b4ac
2 parents ef927db + 2436244
commit fb2b4ac
Show file tree

Hide file tree

Showing 265 changed files with 50,782 additions and 10,946 deletions.
diff --git a/azure-pipelines.yml b/azure-pipelines.yml
@@ -16,7 +16,7 @@ jobs:
       matrix:
         avx2:
           SIMD: 'AVX2'
-          STATIC: 1
+          STATIC: 0
           MPI: 0
           BUILD_TYPE: RelWithDebInfo
         # sse41:
@@ -26,7 +26,7 @@ jobs:
         #   BUILD_TYPE: RelWithDebInfo
         sse2:
           SIMD: 'SSE2'
-          STATIC: 1
+          STATIC: 0
           MPI: 0
           BUILD_TYPE: RelWithDebInfo
         avx2_mpi:
@@ -57,15 +57,15 @@ jobs:
           export CC=gcc-11 ; export CXX=g++-11
           if [ "${STATIC}" -eq "1" ]; then
               cmake -DCMAKE_BUILD_TYPE=${BUILD_TYPE} -DBUILD_SHARED_LIBS=OFF \
-                -DCMAKE_EXE_LINKER_FLAGS="-static -static-libgcc \
-                -static-libstdc++" -DCMAKE_FIND_LIBRARY_SUFFIXES=".a" \
+                -DPREFER_STATIC=1 -DCMAKE_FIND_LIBRARY_SUFFIXES=".a" \
+                -DCMAKE_FIND_LIBRARY_SUFFIXES=".a" \
                 -DENABLE_WERROR=1 -DHAVE_${SIMD}=1 -DHAVE_MPI=${MPI} ..
           else
               cmake -DHAVE_SANITIZER=1 -DCMAKE_BUILD_TYPE=${BUILD_TYPE} \
                 -DENABLE_WERROR=1 -DHAVE_${SIMD}=1 -DHAVE_MPI=${MPI} ..
           fi
 
-          make -j $(nproc --all)
+          make -j $(nproc --all) VERBOSE=1
         displayName: Build foldseek
       - script: |
           if [ "${SIMD}" = "SSE2" ]; then
@@ -113,8 +113,8 @@ jobs:
           CC=${CPREF}-linux-gnu-gcc CXX=${CPREF}-linux-gnu-g++ \
            cmake -DCMAKE_BUILD_TYPE=${BUILD_TYPE} -DHAVE_TESTS=1 \
             -DBUILD_SHARED_LIBS=OFF \
-            -DCMAKE_EXE_LINKER_FLAGS="-static -static-libgcc \
-            -static-libstdc++" -DCMAKE_FIND_LIBRARY_SUFFIXES=".a" \
+            -DCMAKE_EXE_LINKER_FLAGS="-static -static-libgcc -static-libstdc++" \
+            -DPREFER_STATIC=1 -DCMAKE_FIND_LIBRARY_SUFFIXES=".a" \
             -DRust_CARGO_TARGET=${CPREF}-unknown-linux-gnu \
             -DENABLE_WERROR=1 -DHAVE_${SIMD}=1 ..
           make -j $(nproc --all)

diff --git a/data/CMakeLists.txt b/data/CMakeLists.txt
@@ -13,6 +13,9 @@ set(COMPILED_RESOURCES
         evalue_nn.kerasify
         main.js
         vendor.js.zst
+        makepaddeddb.sh
+        result2structprofile.sh
+        createstructsubdb.sh
         multimersearch.sh
         easymultimersearch.sh
         multimercluster.sh

diff --git a/data/createstructsubdb.sh b/data/createstructsubdb.sh
@@ -0,0 +1,33 @@
+#!/bin/sh -e
+
+LIST="$1"
+IN="$2"
+OUT="$3"
+
+if [ -e "${IN}.dbtype" ]; then
+    # shellcheck disable=SC2086
+    "$MMSEQS" base:createsubdb "${LIST}" "${IN}" "${OUT}" ${CREATESTRUCTSUBDB_PAR} \
+        || fail "createsubdb died"
+fi
+
+if [ -e "${IN}_ss.dbtype" ]; then
+    # shellcheck disable=SC2086
+    "$MMSEQS" base:createsubdb "${LIST}" "${IN}_ss" "${OUT}_ss" ${CREATESTRUCTSUBDB_PAR} \
+        || fail "createsubdb died"
+fi
+
+if [ -e "${IN}_ca.dbtype" ]; then
+    # shellcheck disable=SC2086
+    "$MMSEQS" base:createsubdb "${LIST}" "${IN}_ca" "${OUT}_ca" ${CREATESTRUCTSUBDB_PAR} \
+        || fail "createsubdb died"
+fi
+
+# if [ -e "${IN}_h.dbtype" ]; then
+#     # shellcheck disable=SC2086
+#     "$MMSEQS" base:createsubdb "${LIST}" "${IN}_h" "${OUT}_h" ${CREATESTRUCTSUBDB_PAR} \
+#         || fail "createsubdb died"
+# fi
+
+if [ -e "${OUT}.sh" ]; then
+  rm -f -- "${OUT}.sh"
+fi
diff --git a/data/easystructuresearch.sh b/data/easystructuresearch.sh
@@ -26,6 +26,15 @@ if notExists "${TARGET}.dbtype"; then
             || fail "target createdb died"
     fi
     TARGET="${TMP_PATH}/target"
+
+    if [ -n "${GPU}" ]; then
+        if notExists "${TMP_PATH}/target_pad"; then
+            # shellcheck disable=SC2086
+            "$MMSEQS" makepaddedseqdb "${TMP_PATH}/target" "${TMP_PATH}/target_pad" "${TMP_PATH}/pad_tmp" ${MAKEPADDEDSEQDB_PAR} \
+                || fail "makepaddedseqdb died"
+        fi
+        TARGET="${TMP_PATH}/target_pad"
+    fi
 fi
 
 
@@ -75,6 +84,16 @@ if [ -n "${REMOVE_TMP}" ]; then
             # shellcheck disable=SC2086
             "$MMSEQS" rmdb "${TMP_PATH}/target_ss" ${VERBOSITY}
         fi
+        if [ -f "${TMP_PATH}/target_pad" ]; then
+            # shellcheck disable=SC2086
+            "$MMSEQS" rmdb "${TMP_PATH}/target_pad" ${VERBOSITY}
+            # shellcheck disable=SC2086
+            "$MMSEQS" rmdb "${TMP_PATH}/target_pad_h" ${VERBOSITY}
+            # shellcheck disable=SC2086
+            "$MMSEQS" rmdb "${TMP_PATH}/target_pad_ss" ${VERBOSITY}
+            # shellcheck disable=SC2086
+            "$MMSEQS" rmdb "${TMP_PATH}/target_pad_ss_h" ${VERBOSITY}
+        fi
         if [ -f "${TMP_PATH}/query.dbtype" ]; then
             # shellcheck disable=SC2086
             "$MMSEQS" rmdb "${TMP_PATH}/query" ${VERBOSITY}

diff --git a/data/makepaddeddb.sh b/data/makepaddeddb.sh
@@ -0,0 +1,186 @@
+#!/bin/sh -e
+
+exists() {
+	[ -f "$1" ]
+}
+
+if [ "${CLUSEARCH_PAR}" = 0 ]; then
+    if exists "${IN}_ss.dbtype"; then
+        # shellcheck disable=SC2086
+        "$MMSEQS" lndb "${IN}_h"  "${OUT}_tmp_ss_h" ${VERBOSITY} \
+            || fail "lndb died"
+
+        # shellcheck disable=SC2086
+        "$MMSEQS" lndb "${IN}_ss"  "${OUT}_tmp_ss" ${VERBOSITY} \
+            || fail "lndb died"
+
+        # shellcheck disable=SC2086
+        "$MMSEQS" base:makepaddedseqdb "${OUT}_tmp_ss" "${OUT}_ss" ${MAKEPADDEDSEQDB_PAR} \
+            || fail "mmseqs makepaddedseqdb died"
+
+        # shellcheck disable=SC2086
+        "$MMSEQS" rmdb "${OUT}_tmp_ss" ${VERBOSITY} \
+            || fail "rmdb died"
+
+        # shellcheck disable=SC2086
+        "$MMSEQS" rmdb "${OUT}_tmp_ss_h" ${VERBOSITY} \
+            || fail "rmdb died"
+
+        awk '{ print $3"\t"$1; }' "${OUT}_ss.lookup" > "${OUT}_ss.gpu_mapping1"
+
+
+        if exists "${IN}.dbtype"; then
+            # shellcheck disable=SC2086
+            "$MMSEQS" renamedbkeys "${OUT}_ss.gpu_mapping1" "${IN}" "${OUT}" \
+                --subdb-mode 1 ${THREADS_PAR} \
+                || fail "renamedbkeys died"
+        fi
+
+        if exists "${IN}_ca.dbtype"; then
+            # shellcheck disable=SC2086
+            "$MMSEQS" renamedbkeys "${OUT}_ss.gpu_mapping1" "${IN}_ca" "${OUT}_ca" \
+                --subdb-mode 1 ${THREADS_PAR} \
+                || fail "renamedbkeys died"
+        fi
+
+        if exists "${IN}_h.dbtype"; then
+            # shellcheck disable=SC2086
+            "$MMSEQS" renamedbkeys "${OUT}_ss.gpu_mapping1" "${IN}_h" "${OUT}_h" \
+                --subdb-mode 1 ${THREADS_PAR} \
+                || fail "renamedbkeys died"
+        fi
+
+        rm -f -- "${OUT}.lookup"
+        awk '{print $1"\t"$2"\t"int($3/2)}' "${OUT}_ss.lookup" > "${OUT}.lookup"
+        rm -f -- "${OUT}_ss.gpu_mapping1"
+
+    else
+        if exists "${IN}.dbtype"; then
+            # shellcheck disable=SC2086
+            "$MMSEQS" base:makepaddedseqdb "${IN}" "${OUT}" ${MAKEPADDEDSEQDB_PAR} \
+                || fail "mmseqs makepaddedseqdb died"
+        fi
+
+        if exists "${IN}_ca.dbtype"; then
+            # shellcheck disable=SC2086
+            "$MMSEQS" base:makepaddedseqdb "${IN}_ca" "${OUT}_ca" ${MAKEPADDEDSEQDB_PAR} \
+                || fail "mmseqs makepaddedseqdb died"
+        fi
+    fi
+else
+    if exists "${IN}_ss.dbtype"; then
+        # # shellcheck disable=SC2086
+        "$MMSEQS" lndb "${IN}_h"  "${OUT}_tmp_ss_h" ${VERBOSITY} \
+            || fail "lndb died"
+
+        # shellcheck disable=SC2086
+        "$MMSEQS" lndb "${IN}_ss"  "${OUT}_tmp_ss" ${VERBOSITY} \
+            || fail "lndb died"
+
+        # shellcheck disable=SC2086
+        "$MMSEQS" base:makepaddedseqdb "${OUT}_tmp_ss" "${OUT}_ss" ${MAKEPADDEDSEQDB_PAR} \
+            || fail "mmseqs makepaddedseqdb died"
+
+        # shellcheck disable=SC2086
+        "$MMSEQS" rmdb "${OUT}_tmp_ss" ${VERBOSITY} \
+            || fail "rmdb died"
+
+        # shellcheck disable=SC2086
+        "$MMSEQS" rmdb "${OUT}_tmp_ss_h" ${VERBOSITY} \
+            || fail "rmdb died"
+
+        awk '{ print $3"\t"$1; }' "${OUT}_ss.lookup" > "${OUT}_ss.gpu_mapping1"
+
+        awk 'BEGIN{i=0} FNR==NR{name[$3]=1;print $3"\t"$1;i++; next} {if (!($1 in name)){print $1"\t"i; i++}}' \
+        "${OUT}_ss.lookup" "${IN}.lookup" > "${OUT}_ss.gpu_mapping2"
+
+        if exists "${IN}.dbtype"; then
+            # shellcheck disable=SC2086
+            "$MMSEQS" renamedbkeys "${OUT}_ss.gpu_mapping1" "${IN}" "${OUT}" \
+                --subdb-mode 1 ${THREADS_PAR} \
+                || fail "renamedbkeys died"
+        fi
+
+        if exists "${IN}_ca.dbtype"; then
+            # shellcheck disable=SC2086
+            "$MMSEQS" renamedbkeys "${OUT}_ss.gpu_mapping1" "${IN}_ca" "${OUT}_ca" \
+                --subdb-mode 1 ${THREADS_PAR} \
+                || fail "renamedbkeys died"
+        fi
+
+        if exists "${IN}_h.dbtype"; then
+            # shellcheck disable=SC2086
+            "$MMSEQS" renamedbkeys "${OUT}_ss.gpu_mapping1" "${IN}_h" "${OUT}_h" \
+                --subdb-mode 1 ${THREADS_PAR} \
+                || fail "renamedbkeys died"
+        fi
+
+        # shellcheck disable=SC2086
+        "$MMSEQS" renamedbkeys "${OUT}_ss.gpu_mapping2" "${IN}_seq" "${OUT}_seq" \
+            --subdb-mode 1 ${THREADS_PAR} \
+            || fail "renamedbkeys died"
+
+        # shellcheck disable=SC2086
+        "$MMSEQS" renamedbkeys "${OUT}_ss.gpu_mapping2" "${IN}_seq_ca" "${OUT}_seq_ca"  \
+            --subdb-mode 1 ${THREADS_PAR} \
+            || fail "renamedbkeys died"
+
+        # shellcheck disable=SC2086
+        "$MMSEQS" renamedbkeys "${OUT}_ss.gpu_mapping2" "${IN}_seq_h" "${OUT}_seq_h"  \
+            --subdb-mode 1 ${THREADS_PAR} \
+            || fail "renamedbkeys died"
+
+        # shellcheck disable=SC2086
+        "$MMSEQS" renamedbkeys "${OUT}_ss.gpu_mapping2" "${IN}_seq_ss" "${OUT}_seq_ss"  \
+            --subdb-mode 1 ${THREADS_PAR} \
+            || fail "renamedbkeys died"
+
+        if exists "${IN}_clu.dbtype"; then
+            # shellcheck disable=SC2086
+            "$MMSEQS" filterdb "${IN}_clu" "${OUT}_clutmp" --mapping-file  "${OUT}_ss.gpu_mapping2"  ${VERBOSITY} ${THREADS_PAR} \
+                || fail "filterdb died"
+            # shellcheck disable=SC2086
+            "$MMSEQS" renamedbkeys "${OUT}_ss.gpu_mapping1" "${OUT}_clutmp" "${OUT}_clu"  \
+                --subdb-mode 0 ${THREADS_PAR} \
+                || fail "renamedbkeys died"  
+
+            # shellcheck disable=SC2086
+            "$MMSEQS" rmdb "${OUT}_clutmp" ${VERBOSITY} \
+                || fail "rmdb died"
+        fi
+
+        if exists "${IN}_aln.dbtype"; then
+            # shellcheck disable=SC2086
+            "$MMSEQS" filterdb "${IN}_aln" "${OUT}_alntmp" --mapping-file  "${OUT}_ss.gpu_mapping2"  ${VERBOSITY} ${THREADS_PAR} \
+                || fail "filterdb died"
+            # shellcheck disable=SC2086
+            "$MMSEQS" renamedbkeys "${OUT}_ss.gpu_mapping1" "${OUT}_alntmp" "${OUT}_aln"  \
+                --subdb-mode 0 ${THREADS_PAR} \
+                || fail "renamedbkeys died"  
+            # shellcheck disable=SC2086
+            "$MMSEQS" rmdb "${OUT}_alntmp" ${VERBOSITY} \
+                || fail "rmdb died"
+        fi
+
+        rm -f -- "${OUT}.lookup"
+        awk '{print $1"\t"$2"\t"int($3/2)}' "${OUT}_ss.lookup" > "${OUT}.lookup"
+        rm -f -- "${OUT}_ss.gpu_mapping1"
+        rm -f -- "${OUT}_ss.gpu_mapping2"
+
+        rm "${OUT}_seq"
+        rm "${OUT}_seq_ss"
+        rm "${OUT}_seq_ca"
+        rm "${OUT}_seq_h"
+
+    else
+        if exists "${IN}.dbtype"; then
+            # shellcheck disable=SC2086
+            "$MMSEQS" base:makepaddedseqdb "${IN}" "${OUT}" ${MAKEPADDEDSEQDB_PAR} \
+                || fail "mmseqs makepaddedseqdb died"
+        fi
+    fi
+fi
+
+if [ -e "${OUT}.sh" ]; then
+  rm -f -- "${OUT}.sh"
+fi 
diff --git a/data/result2structprofile.sh b/data/result2structprofile.sh
@@ -0,0 +1,34 @@
+#!/bin/sh -e
+
+IN1="$1"
+IN2="$2"
+RESULT="$3"
+OUT="$4"
+
+if [ -e "${IN1}.dbtype" ]; then
+    # shellcheck disable=SC2086
+    "$MMSEQS" base:result2profile "${IN1}" "${IN2}" "${RESULT}" "${OUT}" ${PROFILE_PAR} \
+        || fail "result2profile died"
+fi
+
+if [ -e "${IN1}_ss.dbtype" ]; then
+    # shellcheck disable=SC2086
+    "$MMSEQS" base:result2profile "${IN1}_ss" "${IN2}_ss" "${RESULT}" "${OUT}_ss" ${PROFILE_SS_PAR} \
+        || fail "result2profile died"
+fi
+
+if [ -e "${IN1}_ca.dbtype" ]; then
+    # shellcheck disable=SC2086
+    "$MMSEQS" lndb "${IN1}_ca"  "${OUT}_ca" ${VERBOSITY} \
+        || fail "Create lndb died"
+fi
+
+if [ -e "${IN1}_h.dbtype" ]; then
+    # shellcheck disable=SC2086
+    "$MMSEQS" lndb "${IN1}_h"  "${OUT}_h" ${VERBOSITY} \
+        || fail "Create lndb died"
+fi
+
+if [ -e "${OUT}.sh" ]; then
+  rm -f -- "${OUT}.sh"
+fi