Hi,

I get an error when loading a Metrichor generated fast5 file set in IONiseR.

Here is an example on one file :

> readFast5Summary(c("example.fast5"))
Checking file validity
Reading Channel Data
Reading Raw Data
Reading Template Data
Error in H5Aopen(gid, "num_events") :
  HDF5. Attribute. Unable to initialize object.

The file can be downloaded :

http://genoweb.toulouse.inra.fr/~klopp/Minion/example.fast5

> sessionInfo()
R version 3.3.0 (2016-05-03)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: CentOS release 6.5 (Final)http://genoweb.toulouse.inra.fr/~klopp/Minion/example.fast5

locale:
[1] C

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
[1] gridExtra_2.2.1 ggplot2_2.1.0   IONiseR_1.2.3  

loaded via a namespace (and not attached):
 [1] Rcpp_0.12.5                magrittr_1.5              
 [3] XVector_0.12.0             GenomicRanges_1.24.2      
 [5] BiocGenerics_0.18.0        zlibbioc_1.18.0           
 [7] GenomicAlignments_1.8.3    IRanges_2.6.1             
 [9] munsell_0.4.3              BiocParallel_1.6.2        
[11] lattice_0.20-33            colorspace_1.2-6          
[13] R6_2.1.2                   dplyr_0.5.0               
[15] hwriter_1.3.2              plyr_1.8.4                
[17] GenomeInfoDb_1.8.1         tools_3.3.0               
[19] SummarizedExperiment_1.2.3 parallel_3.3.0            
[21] grid_3.3.0                 rhdf5_2.16.0              
[23] Biobase_2.32.0             data.table_1.9.6          
[25] gtable_0.2.0               DBI_0.4-1                 
[27] latticeExtra_0.6-28        lazyeval_0.2.0            
[29] assertthat_0.1             tibble_1.0                
[31] tidyr_0.6.0                RColorBrewer_1.1-2        
[33] S4Vectors_0.10.1           bitops_1.0-6              
[35] scales_0.4.0               Biostrings_2.40.2         
[37] Rsamtools_1.24.0           stats4_3.3.0              
[39] ShortRead_1.30.0           chron_2.3-47             

Your assistance is much appreciated.

Christophe

Dear Christophe,

Thanks for your question.  Providing access to the fast5 file is very helpful.  I've got a couple of questions that may help get IONiseR into a state that can read it.  Looking at it's contents, it formatted in a subtle different way to any to any fast5 files I've encountered before (this is not uncommon, ONT change the format regularly!).  

It also looks like you may have processed this twice, maybe using the MinKNOW live base calling, and then the Metrichor cloud-based base caller.  That might explain why the structure is more complex than files I've seen previously.

From the internal file name, I also wonder if this was generated during the Mux scan, which isn't really designed to produced results and more to calibrate the device.  Trying to analyse data from this step wouldn't really be meaningful and it may be that the fast5 files produced have a non-standard format.

If either of these things are true and the file come from some 'unusual' workflow then I suggest continuing with the experiment and let me know if IONiseR fails to cope with your next set of fast5 files.  However, if this was produced with the standard workflow for 'normal' data then I'll patch IONiseR to cope with ONTs ever shifting format.