Troubleshooting

Report any troubleshooting in this page

salmon indexing

When indexing a reference with salmon, one first needs to generate a transcript fasta file. By default, this step (performed with gffread) may lead to warnings and errors when calling salmon index. The warnings and errors may be related to transcripts being shorter than the kmer length, or duplicated entries. Using the -J option in gffread should solve the problem; this option removes the transcript that have no start or no stop.

star segmentation fault

Most of the time, when errors occurs at star level, this is a segmentation fault, which is sometimes difficult to identify. We noticed that this usually happens with small genomes. For small genome, we build the index using the recommended value from star manual. Yet, this may lead to segmentation fault. In such case, you will currently need to build the indexing manually. Imagine that your genome is in GENOMES/MYFAVORITE/ then, the star indexing is saved into GENOMES/MYFAVORITE/star asuming your fasta is stored in GENOMES/MYFAVORITE/MYFAVORITE.fa. So you will need to type::

STAR --runMode genomeGenerate --genomeFastaFiles GENOMES/MYFAVORITE/MYFAVORITE.fa \
    --genomeDir GENOMES/MYFAVORITE/star --genomeSAindexNbases 9

Here, by default the genomeSAindexNbases may not be the proper one to use. THis seems to be a known issue. For example, for ecoli, our default value is 10 (according to the STAR manual spec) and yet, we had segmentation fault. Setting the value to 9 remove the seg fault.