Installation error: `memcpy@GLIBC_2.14'

[Jidgdoi]

Hi, I'm trying to install _rmats_turbo_v4_12 on a server running on CentOS 7.

make[1]: Entering directory '/beegfs/data/work/shared/xxxxx/bin/rmats_turbo_v4_1_2/rMATS_C'
/work/shared/xxxxx/bin/rmats_turbo_v4_1_2/conda_envs/rmats/bin/x86_64-conda-linux-gnu-cc -I/work/shared/xxxxx/bin/rmats_turbo_v4_1_2/conda_envs/rmats/include -Wall -O2 -msse2 -funroll-loops -fopenmp -o rMATSexe src/main.c src/myfunc.c src/util.c lbfgs_scipy/lbfgsb.o lbfgs_scipy/linpack.o lbfgs_scipy/timer.o -L/work/shared/xxxxx/bin/rmats_turbo_v4_1_2/conda_envs/rmats/lib -lgsl -lgslcblas -lm -lm -lgfortran -lgsl -lgslcblas -lgomp -lblas -llapack
/beegfs/data/work/shared/xxxxx/bin/rmats_turbo_v4_1_2/conda_envs/rmats/bin/../lib/gcc/x86_64-conda-linux-gnu/9.5.0/../../../../x86_64-conda-linux-gnu/bin/ld: /work/shared/xxxxx/bin/miniconda3/lib/libquadmath.so.0: undefined reference to `memcpy@GLIBC_2.14'
collect2: error: ld returned 1 exit status
make[1]: *** [Makefile:40: rMATSexe] Error 1
make[1]: Leaving directory '/beegfs/data/work/shared/xxxxx/bin/rmats_turbo_v4_1_2/rMATS_C'
make: *** [Makefile:6: build] Error 2

and am facing this issue undefined reference to memcpy@GLIBC_2.14.

I tried the solution from StackOverflow https://stackoverflow.com/questions/12286460/undefined-reference-to-mempcyglibc-2-14-when-compiling-on-linux by adding the code line to rMATC_C/src/version.c. Probably did wrong but it didn't change anything.

Any suggestions ?

[EricKutschera]

It looks like you are using conda. If you don't need to build from the source code then you could try installing the rmats v4.1.2 conda package:

conda install -c conda-forge -c bioconda rmats=4.1.2

For the actual error message /work/shared/xxxxx/bin/miniconda3/lib/libquadmath.so.0: undefined reference to 'memcpy@GLIBC_2.14', it's saying that some library requires something that should be in glibc v2.14 or later. You could try updating the version of glibc on your system

[docmab23]

Is rmats v4.1.2 a version fo rmats_turbo or the standard rMats , also generally how much time would the pipeline take (pre and post step together), running it on 18 threads, there are 4 samples (2 reps in each group), the STAR step was significantly faster, but the getting the .rmats files is pretty slow, for my previous run (on a different dataset) it took me like ~13 hours.

Thanks

[EricKutschera]

All versions like v4.x.x are rMATS-turbo

rMATS can process each input sample (bam file) on a separate thread in the prep and post steps. If you only have 4 samples then only 4 threads will be used for those steps. For each sample (thread), the prep step would maybe take 1 hour per 200 million alignments. The post step depends on the events detected so it is harder to estimate. Usually the post step would only take a few hours unless --novelSS is used which can lead to very long running time for large datasets