Hi Alan, "Alan O'Cais" <alan.oc...@cecam.org> writes:
> Hmm, the patch for that > (https://github.com/easybuilders/easybuild-easyconfigs/blob/develop/easybuild/easyconfigs/m/medaka/medaka-1.4.3_use-SAMtools-dep.patch) > seems to have been dropped in 1.5.0. > > Bringing that patch back may fix your problem. Thanks for the pointer. I was able to port the patch and that did indeed fix my problem. However, my next problem is that medaka 1.6.0 needs TensorFlow 2.7.0. Unfortunately there are no ECs beyond 2.6.0, even though the current version is 2.9.0 (ML on HPC is not going out of fashion, is it?). I see Sam has put in a PR #14990 for 2.7.0, but for foss/2021b, so I guess I'll have to look into bumping my EC up to that toolchain. Cheers, Loris > Alan > > On Tue, 17 May 2022 at 15:41, Alan O'Cais <alan.oc...@cecam.org> wrote: > > Seems like Medaka is doing naughty stuff: > https://github.com/nanoporetech/medaka/blob/master/Makefile#L76 > > You probably need to patch the Makefile so an EB-provided version of those > downloaded deps are used. > > On Tue, 17 May 2022 at 15:27, Loris Bennett <loris.benn...@fu-berlin.de> > wrote: > > Hi, > > I have successfully installed > > medaka-1.5.0-foss-2021a.eb > > I then created an EC with the following changes for medaka 1.6.0: > > $ diff > /trinity/shared/easybuild/software/EasyBuild/4.5.4/easybuild/easyconfigs/m/medaka/medaka-1.5.0-foss-2021a.eb > medaka-1.6.0-foss-2021a.eb > 13c13 > < version = '1.5.0' > --- > > version = '1.6.0' > 26c26 > < ('TensorFlow', '2.5.3'), > --- > > ('TensorFlow', '2.6.0'), > 35c35 > < ('ont-fast5-api', '4.0.0'), > --- > > ('ont-fast5-api', '4.0.2'), > 47c47 > < 'checksums': > ['8a0b8d5997fe45b8d5a1317b2b6bfe117cb6b551928faef62e4e958cf14b5ea4'], > --- > > 'checksums': > ['a8389f90cba77095061c4ff22c7cce8d73908e2451fa1d07b94c2f8ce4152628'], > > The new TensorFlow and ont-fast5-api packages seem to be built and > installed correctly, but building medaka itself fails with the following > error: > > ld.gold: error: libhts.a(hts.o): requires unsupported dynamic reloc 11; > recompile with -fPIC > > Does anyone have an idea what might be causing this? > > Cheers, > > Loris > > -- > Dr. Loris Bennett (Herr/Mr) > ZEDAT, Freie Universität Berlin Email loris.benn...@fu-berlin.de > -- Dr. Loris Bennett (Herr/Mr) ZEDAT, Freie Universität Berlin Email loris.benn...@fu-berlin.de
