memory usage hisat2-align died with signal 9 (KILL)

[murheisa]

Hello, I´ve been trying to use hisat2 to align the host genome to metatranscriptomics sequences.
In this case, the host genome is 3gb and I´ve tried to run hisat2 in a cluster with the following code.

hisat2 -x hisat2_index
-1 /fwd.fq.gz
-2 /rev.fq.gz
--very-sensitive
--threads 20
--un-conc-gz non_host.fq.gz
-S host_mapped.sam

I´ve tried different number of threads: 4,10, 15,20 and 28 and in any case the ram memory scales to more than 125gb and 256gb.
The sam file reached 40gb and the non host files are 70Mb. The error file says the following:
Time loading forward index: 00:00:02
Time loading reference: 00:00:01
(ERR): hisat2-align died with signal 9 (KILL)

Please is there any way I can improve the memory issues?

[ydelen2]

I have encountered exactly same issue with hisat2, but it could not be resolved, no matter how much memory you use. Using more threads is not the solution, because more threads gonna use more RAM, but you might not get the process completed. Actually hisat2 tends to map reads to multiple sites in the reference, which takes much more time and memory, not efficient. I used STAR for the analysis, which completed mapping same 58 GB of read file (each in paired end) only under 32 GBs of RAM. trying using STAR.

[mourisl]

Sorry, I missed the issue. Which index file did you use? Is it from pre-built or you created your own index? Could you please also run the hisat with the option "--verbose" that paste the running log here? Thank you!