BamUtil: FAQ

From Genome Analysis Wiki
Jump to: navigation, search

For problems with libStatGen, see LibStatGen Troubleshooting

Compile Problems

HashErrorModel.cpp:96: error: ‘class std::unordered_map, std::equal_to, std::allocator > >’ has no member named ‘at’

Error:
HashErrorModel.cpp: In member function ‘uint8_t HashErrorModel::getQemp(BaseData&)’:
HashErrorModel.cpp:96: error: ‘class std::unordered_map, std::equal_to, std::allocator > >’ has no member named ‘at’
make[2]: *** [../obj/HashErrorModel.o] Error 1
HashErrorModel by default uses an unordered_map which is only found in C++11/C++0x.
The code attempts to automatically detect whether or not you have C++11/C++0x by checking if the gcc version > X4.3.0.
If you see this failure, that means this check did not properly work and you will need to manually disable the C++11/C++0x code. There are a couple of ways to do this:
  1. Update bamUtil/Makefile.inc, line 7, change CXX11_AVAIL ?= 1 to CXX11_AVAIL = 0
  2. Set CXX11_AVAIL to 0 on the command line:
    make CXX11_AVAIL=0
  3. Set CXX11_AVAIL = 0 in your environment

By making one of these changes, the code should compile and use map instead of unordered_map.

ld: lto: could not merge in ../obj/BamExecutable.o because Unknown instruction for architecture x86_64

ERROR:
ld: lto: could not merge in ../obj/BamExecutable.o because Unknown instruction for architecture x86_64
collect2: ld returned 1 exit status
make[2]: *** [../bin/bam] Error 1
make[1]: *** [src] Error 2
make: *** [bamUtil/] Error 2
This error was seen when running on Mac OS X 10.7.5 on a 64-bit kernel with gcc/g++: i686-apple-darwin11-llvm-g++-4.2 (GCC) 4.2.1 (Based on Apple Inc. build 5658) (LLVM build 2336.1.00)
The error appears to be due to the default build optimization: -O4
If you see this error, try compiling for debug (optimization -O0): make debug
The debug executable is: bamUtil/bin/debug/bam
If the debug compile worked and you would prefer an optimized executable (runs a little faster), then update the optimization flag to -O3:
  1. Open libStatGen/Makefiles/Makefile.include
  2. Change OPTFLAG_OPT?=-O4 to OPTFLAG_OPT?=-O3
  3. Run make clean; make in both libStatGen and bamUtil
  4. The optimized executable is bamUtil/bin/bam


BamUtil: stats

Overflow on the pileup buffer: specifiedPosition = ####, pileup buffer start position: ####, pileup buffer end position: ####

 Exiting due to ERROR:
	Overflow on the pileup buffer: specifiedPosition = 93252, pileup buffer start position: 92227, pileup buffer end position: 93251
By default, bam stats assumes that a single read covers less than 1024 reference bases.
This type of error appears if a read is longer than that. This is most likely to happen if you have large skipped regions in your CIGARs ('N's).
You need to increase the size of the pileup buffer to cover the largest number of reference bases a single read covers.
To fix this problem, use the --bufferSize 3000 parameter, replacing 3000 by the appropriate number to handle the largest size of the reference covered by a read. You can increase this to a large number - it will just take up more memory.
Future versions of bamUtil will print an additional error message including the recordName and CIGAR of the record that failed. You can use the CIGAR to come up with a better setting for --bufferSize, although keep in mind that future records could cover even a larger number of bases than that failing record.
To print all of the Cigars in the file that contain an 'N' (skip), you can use (replace test.bam with your bam file):
  ../bin/bam findCigars --noph --cskip --in test.bam --out - |grep -v "^@" |cut -f 6
This may help to determine a setting for --bufferSize.