Getting command line bioinformatics tools working on Android - part 2

In the previous post, I showed how Linux command line bioinformatics tools (such as minimap2) can be run on an Android mobile phone through Android Debug Bridge. That method required us to issue commands to the phone from the host PC via USB. In this post, I will show how we can make it a bit fancier, by issuing commands directly from the mobile phone. In summary, we will install a virtual terminal app to the phone and issue commands from there.

Read More

Getting command line bioinformatics tools working on Android

Once my laptop completely died and did not have much to do until the new laptop arrived. But I had my Android mobile phone and wondered how hard will it be to get some common ONT bioinformatics tools working there. On that day, I with my friend @danielltb realised that is not that hard (but in a very hacky way). Here are the steps, if anyone is crazy enough to attempt it. Interestingly, this method does NOT require a rooted phone1.

  1. At the time of writing Android (tested on Android 7 and 8) seem to allow executing binaries from /data/local/tmp through the Android Debug Bridge (ADB). As long as this is not blocked in the future versions, the method should work. 

Read More

Generating portable binaries for ONT tools

Compiling software can sometimes be a nightmare due to numerous dependencies. This is specifically the case for bioinformatics tools that utilise signal level data from Oxford Nanopore (ONT) sequencers. According to my experience, the major cause behind compilation troubles in ONT tools is the Hierarchical Data Format 5 (HDF5) library1. While a system admin may enjoy tedious compilations, it is not the case for users of bioinformatics tools. What if the tool developers release pre-compiled binaries? Some would object this as it is not a perfect solution. Nevertheless, I believe that it is far better than releasing an unusable tool due to users giving up at the compilation stage. Further, pre-compiled binaries are less bulky compared to docker images. Generating a “portable binary” that runs on numerous Linux distributions/version is tricky, but possible with some additional work from the developer’s side.

  1. Currently, the raw signal data from the ONT sequencers are stored in HDF5 file format. Possibly due to the complexity of HD5, there are no alternate library implementations than the official library from the HDF Group. Compiling the HDF5 library takes time. Luckily, package managers’ versions of HDF5 library exists, but there seem to be some inconsistencies across various distributions. Thus, a software developed on one Linux distribution will rarely compile without any trouble on a different system. For example, the header file resides directly *include* directory on certain systems, while in some other system it can be <hdf5/hdf5.h> or <hdf5/serial/hdf5.h>. 

Read More