This is a test version of Biostars. For the public version, visit https://www.biostars.org.
[E::bwa_idx_load_from_disk] fail to locate the index files

I'm really struggling to get my bwa (version 0.7.15) to find my index.

My script for indexing was this

bwa index -p ref_genome -a bwtsw $HOME/hg19.fa

which generated the following files in the ref_genome directory:

ref_genome.amb  ref_genome.ann  ref_genome.bwt  ref_genome.pac  ref_genome.sa

I then enter the following to align to my reference

bwa mem -t 4 /rds/general/user/snw02/home/bwa_index/ref_genome ~/trimmed_reads/paired_end/s1_S1100619_1P.fq.gz ~/trimmed_reads/paired_end/s1_S1100619_2P.fq.gz >  s1_S1100619_aln_pe.sam

and I'm still getting the error message above (I moved ref_genome into another directory called bwa_index to see if it would help. It didn't!)

Any suggestions?

bwa

Are all five files in /rds/general/user/snw02/home/bwa_index/? If so this is unusual. Perhaps you encountered an error when creating that initial index. Did you check the log for that?

Are the sizes of your ref_genome files in this neighborhood?

64K     genome.fa.amb
64K     genome.fa.ann
3.5G    genome.fa.bwt
890M    genome.fa.pac
1.8G    genome.fa.sa

Similar...they are:

-rw------- 1 snw02 hpc-gpt30 17K Mar 17 17:24 ref_genome.amb
-rw------- 1 snw02 hpc-gpt30 7.9K Mar 17 17:24 ref_genome.ann
-rw------- 1 snw02 hpc-gpt30 5.9G Mar 17 17:23 ref_genome.bwt
-rw------- 1 snw02 hpc-gpt30 1.5G Mar 17 17:24 ref_genome.pac
-rw------- 1 snw02 hpc-gpt30 3.0G Mar 17 18:13 ref_genome.sa

I'm wondering if it is because I didn't create the original reference with a .fa extension?

I also moved it from the temporary directory (where everything is done on the cluster) to my home directory.

I think I need to probably just create the index again don't I???

1 answer

It appears that only snw02 account has read permissions for these files. Are you running the alignments from this account? Is your home directory available on nodes where the actual job runs. On some clusters home directories may not be available on work nodes so leaving these files in a directory that is accessible on cluster work nodes should fix this problem.

Yeah I moved the files to the nodes where the jobs run and it works fine now!

Log in to answer this question.