giraffe_accuracy_primary_kubernetes

(kubectl delete job xhchang-giraffe-primary || true) && kubectl apply -f - <<'EOF'
apiVersion: batch/v1
kind: Job
metadata:
  name: xhchang-giraffe-primary
spec:
  ttlSecondsAfterFinished: 86400
  template:
    spec:
      containers:
      - name: xhchang-giraffe-primary
        imagePullPolicy: Always
        image: xhchang/vg:giraffe-paper
        command:
        - /bin/bash
        - -c
        - |
          set -ex
          cd /tmp
          printf "graph\tgbwt\treads\tpairing\tspeed\tcorrect\tmapq60\twrong_mapq60\tscore\n" > report.tsv
          printf "correct\tmq\tscore\taligner\n" > roc_stats.tsv
          for SPECIES in human  ; do
              case "${SPECIES}" in
              yeast)
                  GRAPHS=(S288C)
                  READSETS=(DBVPG6044 DBVPG6765 N44 UWOPS034614 UWOPS919171 Y12 YPS138)
                  GBWTS=(raw)
                  ;;
              human)
                  GRAPHS=(1000gp hgsvc)
                  READSETS=(novaseq6000 hiseqxten hiseq2500)
                  # We really only need one GBWT because we have no variation
                  GBWTS=(cover)
                  ;;
              esac
              for GRAPH in ${GRAPHS[@]} ; do
                  for READS in ${READSETS[@]} ; do
                      case ${GRAPH} in
                      1000gp)
                          GRAPH_BASE=s3://vg-k8s/profiling/graphs/v2-2/generic/primary/hs38d1/primaryhs38d1
                          aws s3 cp s3://vg-k8s/profiling/reads/sim/for-NA19239/1000gp/hs38d1/liftover_nosegdups/${READS}/out_sim_gbwt/sim.gam ./sim.gam
                          ;;
                      hgsvc)
                          GRAPH_BASE=s3://vg-k8s/profiling/graphs/v2-2/generic/primary/hs38d1/primaryhs38d1
                          aws s3 cp s3://vg-k8s/profiling/reads/sim/for-NA19240/hgsvc/grch38/${READS}/out_sim_gbwt/sim.gam ./sim.gam
                          ;;
                      S288C)
                          GRAPH_BASE=s3://vg-k8s/profiling/graphs/v2/generic/primary/S288C/primaryS288C
                          aws s3 cp s3://vg-k8s/profiling/reads/sim/yeast/sim-${READS}.gam ./sim.gam
                          ;;
                      esac
                      aws s3 cp ${GRAPH_BASE}.xg ./graph.xg
                      aws s3 cp ${GRAPH_BASE}.dist ./graph.dist
                      for GBWT in ${GBWTS[@]} ; do
                          rm -f graph.gbwt
                          rm -f graph.gg
                          rm -f graph.min
                          if [[ "${GBWT}" == "raw" ]] ; then
                              aws s3 cp ${GRAPH_BASE}.gbwt ./graph.gbwt
                              aws s3 cp ${GRAPH_BASE}.gg ./graph.gg
                              aws s3 cp ${GRAPH_BASE}.min ./graph.min
                          else
                              aws s3 cp ${GRAPH_BASE}.${GBWT}.gbwt ./graph.gbwt
                              aws s3 cp ${GRAPH_BASE}.${GBWT}.gg ./graph.gg
                              aws s3 cp ${GRAPH_BASE}.${GBWT}.min ./graph.min
                          fi
                          for PAIRING in single paired ; do 
                              if [[ ${PAIRING} == "paired" ]] ; then
                                  PAIRED="-i"
                              elif [[ ${PAIRING} == "single" ]] ; then 
                                  PAIRED=""
                              fi
                              SPEED="$(vg giraffe -x graph.xg -H graph.gbwt -g graph.gg -m graph.min -d graph.dist -G sim.gam ${PAIRED} -t 22 -p 2>&1 >mapped.gam | grep speed | sed 's/[^0-9\.]//g')"
                              vg gamcompare -r 100 -s <(vg annotate -m -x graph.xg -a mapped.gam) sim.gam 2>count | vg view -aj - > compared.json
                              CORRECT_COUNT="$(sed -n '1p' count | sed 's/[^0-9]//g')"
                              SCORE="$(sed -n '2p' count | sed 's/[^0-9\.]//g')"
                              MAPQ="$(grep mapping_quality\":\ 60 compared.json | wc -l)"
                              MAPQ60="$(grep -v correctly_mapped compared.json | grep mapping_quality\":\ 60 | wc -l)"
                              IDENTITY="$(jq '.identity' compared.json | awk '{sum+=$1} END {print sum/NR}')"
                              echo ${GRAPH} ${GBWT} ${READS} ${PAIRING} ${SPEED} ${CORRECT_COUNT} ${MAPQ} ${MAPQ60} ${IDENTITY} ${SCORE}
                              printf "${GRAPH}\t${GBWT}\t${READS}\t${PAIRING}\t${SPEED}\t${CORRECT_COUNT}\t${MAPQ}\t${MAPQ60}\t${SCORE}\n" >> report.tsv

                              jq -r '(if .correctly_mapped then 1 else 0 end|tostring) + "," + (.mapping_quality|tostring) + "," + (.score|tostring)' compared.json | sed 's/,/\t/g' | sed "s/$/\tgiraffe_primary_${GRAPH}${GBWT}${READS}${PAIRING}/" >> roc_stats.tsv
                          done
                      done
                  done
              done
          done
          sed -i 's/single//g ; s/paired/-pe/g ; s/null/0/g' roc_stats.tsv
          gzip roc_stats.tsv
          aws s3 cp report.tsv s3://vg-k8s/users/xhchang/giraffe_experiments_2/liftover/nosegdups/report_giraffe_primary.tsv
          aws s3 cp roc_stats.tsv.gz s3://vg-k8s/users/xhchang/giraffe_experiments_2/liftover/nosegdups/roc_stats_giraffe_primary.tsv.gz
        volumeMounts:
        - mountPath: /tmp
          name: scratch-volume
        - mountPath: /root/.aws
          name: s3-credentials
        resources:
          requests:
            cpu: 24
            memory: "120Gi"
            ephemeral-storage: "150Gi"
          limits:
            cpu: 24
            memory: "120Gi"
            ephemeral-storage: "150Gi"
      restartPolicy: Never
      volumes:
      - name: scratch-volume
        emptyDir: {}
      - name: s3-credentials
        secret:
          secretName: shared-s3-credentials
  backoffLimit: 0
EOF