hisat2-build indexing produces more than 8 output files

Question

[ddowlin]

Hi all,

I am trying to assemble a primate transcriptome using hisat2/stringtie. First I want to index the genome using hisat2. I used both --ss and --exon to provide information on splice sites and exons.

The manual states that 8 files should be produced after indexing 1.ht2 to 8.h2t. However, I only have six files with this suffix (5.ht2 and 6.ht2 are missing) and one file is completely empty. Additionally I have 20 .rf files (0.rf to 19.rf).

Does anyone know why this is? Was there an error with indexing or can I use these files as is?

[Vasu]

You can download pre-built indexes from the Hisat2 website [https://ccb.jhu.edu/software/hisat2/index.shtml] u may also download genome_snp, genome_tran and / or genome_snp_tran.

[ddowlin]

Thanks for the suggestion. Unfortunately pre-made indices aren't available for the species I am interested in.

[Satyajeet Khare]

Can you try without the --ss and --exon options? These options need big RAM. If its a big genome, indexing may have stopped with error.

Best,

[ddowlin]

Thanks--I tried without the --ss and --exon options and it seems to have worked OK.

[Satyajeet Khare]

Yup. So my guess is that your indexing may have stopped due to RAM issues. You can still provide the 'splice sites' file at the time of alignment.

Best

[GenoMax]

If there was no error produced during index build then go ahead and use them. Programs often will store these indexes in formats they choose/like. If there was an error then post that here.

[psm]

I know this post is old but just thought I'd reply to this specific comment since I struggled with this issue for a while - I think HISAT2 is supposed to produce exactly 8 files, and as Satyajeet Khare mentioned may be the most correct explanation. I also got more than 8 files every time my process terminated prematurely due to RAM issues. I think the .rf files are temporary files and should NOT be used for alignment.

[GenoMax]

Do you know if your build run log had something in it that indicated a problem when you had the truncation happen? If HISAT2 devs have not done their due diligence to flag that in the log output then that is not a good thing.

Were you able to align data to (with what may be truncated index files)? One would expect HISAT2 to throw an error, if it detects an incomplete/corrupt index set.

Edit: hisat2-build manual says the following, which is incorrect. There should be 8 files in properly built indexes.

hisat2-build outputs a set of 6 files with suffixes .1.ht2, .2.ht2, .3.ht2, .4.ht2, .5.ht2, .6.ht2, .7.ht2, and .8.ht2.

[gaoyufeng1109]

Hi all, I indexed the genome using hisat2 and used both --ss and --exon to provide information on splice sites and exons. And I also met the same problem. I also only got six files with this suffix (5.ht2 and 6.ht2 are missing). Have you solve the problem? And do you know the reason?

[GenoMax]

Have you tried to use the index for an alignment? Don't go on the number of files produced. Properly built indexes should have 8 files.

[gaoyufeng1109]

Thanks for your help. I have not mapped the reads. OK. I will try.

[psm]

I am having a similar issue - many .rf files but less than 8 .ht2 files. Was this ever resolved? Are they okay to use?

[MK]

I doubt it is ok to use them. I have the same problem. Only 4 files are generated. I tried to use them for alignment but it won't work.