run_wga_gpu crashes on all-against-all human chimp alignment about segalign

[glennhickey]

I'm seeing another crash when I put the masked human/chimp/gorilla data through cactus.
The command that fails is

run_wga_gpu tmp_fv6o6ah.tmp tmp_fv6o6ah.tmp --format=cigar --notrivial --step=3 --ambiguous=iupac100100 --ydrop=3000

The file is s3://glennhickey/share/tmp_fv6o6ah.tmp
The output ends with:

       Chromosome block 11 interval 24/28 (230000000:240000000) with buffer 1
        Chromosome block 11 interval 25/28 (240000000:250000000) with buffer 1
        Chromosome block 11 interval 26/28 (250000000:260000000) with buffer 1
        Chromosome block 11 interval 27/28 (260000000:270000000) with buffer 1
        Chromosome block 11 interval 28/28 (270000000:275805167) with buffer 1

        real    15m46.438s
        user    584m26.216s
        sys     186m20.207s

        Error in LASTZ process!
        truncating alignment starting at (1272489,1179798);  anchor at (2041888,1949197)
        truncation can be reduced by using --allocate:traceback to increase traceback memory
        truncating alignment ending at (1231827,1139136);  anchor at (462539,369848)
        truncating alignment starting at (3052559,1179798);  anchor at (3821958,1949197)
        truncation can be reduced by using --allocate:traceback to increase traceback memory
        truncating alignment ending at (3011897,1139136);  anchor at (2242609,369848)
        truncating alignment starting at (17191591,2729808);  anchor at (17960967,3499184)
        truncation can be reduced by using --allocate:traceback to increase traceback memory
        truncating alignment ending at (18730380,4268597);  anchor at (17960967,3499184)
        truncating alignment ending at (16515980,2054165);  anchor at (15746588,1284773)
        truncating alignment ending at (124854312,68899144);  anchor at (124086556,68131126)
        truncation can be reduced by using --allocate:traceback to increase traceback memory
        truncating alignment starting at (3045081,2842260);  anchor at (3814457,3611636)
        truncation can be reduced by using --allocate:traceback to increase traceback memory
        truncating alignment ending at (4583870,4381049);  anchor at (3814457,3611636)
        truncating alignment ending at (2369438,2166649);  anchor at (1600046,1397257)
        truncating alignment starting at (249243252,224227);  anchor at (250012619,993594)
        truncation can be reduced by using --allocate:traceback to increase traceback memory
        truncating alignment ending at (17982972,93954331);  anchor at (17213581,93184940)
        truncation can be reduced by using --allocate:traceback to increase traceback memory
        truncating alignment starting at (17138306,93109665);  anchor at (17907673,93879032)
        truncation can be reduced by using --allocate:traceback to increase traceback memory

The cactus command run just for the record was

/usr/bin/time -v cactus-blast ./js-no-realign ./trio.txt ./blast-out-gpu --configFile cactus/config-gpu.xml --realTimeLogging --logFile log-gpu-no-realign.txt --binariesMode local --root hc --cleanWorkDir never

cat ../trio.txt
((hg38:0.01, panTro6:0.01)hc:0.11,gorGor5:0.01);

hg38 /home/ubuntu/Homo_sapiens.fa
panTro6 /home/ubuntu/Pan_troglodytes.fa
gorGor5 /home/ubuntu/Gorilla_gorilla.fa

[gsneha26]

run-wga-gpu changed accordingly to only fail on LASTZ errors not warnings.