psites_number table is not generated with v1.2.15

[hsinyenwu]

I think I still encounter an issue that is similar to issue#32
i.e., I did not get the psites_number table in the hdf5 file.

  group           name       otype dclass   dim
0     /        p_sites H5I_DATASET   VLEN 46826
1     / transcript_ids H5I_DATASET STRING 46826

Also the dimension of the p_sites table seems to be wrong.
Otherwise the code ran okay and generated other files.

This job is running on skl-119 on Wed May 17 23:21:46 EDT 2023
0% has finished! ^M2% has finished! ^M4% has finished! ^M6% has finished! ^M9% has finished! ^M11% has finished! ^M13% has finished! ^M15% has finished! ^M17% has finished! ^M19% has finished! ^M21% has finished! ^M23% has finished! ^M26% has finished! ^M28% has finished! ^M30% has finished! ^M32% has finished! ^M34% has finished! ^M36% has finished! ^M38% has finished! ^M41% has finished! ^M43% has finished! ^M45% has finished! ^M47% has finished! ^M49% has finished! ^M51% has finished! ^M53% has finished! ^M56% has finished! ^M58% has finished! ^M60% has finished! ^M62% has finished! ^M64% has finished! ^M66% has finished! ^M68% has finished! ^M70% has finished! ^M73% has finished! ^M75% has finished! ^M77% has finished! ^M79% has finished! ^M81% has finished! ^M83% has finished! ^M85% has finished! ^M88% has finished! ^M90% has finished! ^M92% has finished! ^M94% has finished! ^M96% has finished! ^M98% has finished! ^M[2023-05-17 23:29:14] Finished!
        Loading transcripts.pickle ...
        Reading bam file: /mnt/home/larrywu/CTRL_arabidopsis/data/RiboCode_STAR/ribo_mapped/D1//star_D1_Aligned.toTranscriptome.out.bam......
Finished reading bam file!

Any suggestions for how to deal with this issue? Thanks!

[Roleren]

An easy hack is to add it, very easy in R at least:

library(rhdf5)
p_sites <- rhdf5::h5read(h5file_path, "p_sites") # Read in the p_sites object, it can be quite big, ~ 1GB for human  
p_sites_sum <- sum(sapply(p_sites, sum)) # If object is VLEN (i.e. R list), pick this one
p_sites_sum <- as.integer(sum(rowSums(p_sites))) # If object is matrix (i.e. R matrix), pick this one
rhdf5::h5writeDataset(obj = p_sites_sum, h5loc = H5Fopen(h5file),
                        name = "psites_number")

You can also insert the psites directly, object if you want to skip the whole bam creation and hdf5 step.
# p_sites as VLEN, this is much simpler, but you need 1TB memory for large species:
library(hdf5r)
h5file_link <- H5File$new(h5file, mode = "w")
h5file_link[["p_sites"]] <- psites # I usually use as(IntegerList(ORFik::coveragePerTiling(tx, RPF)), "list) on a bigwig or covRle object
# Then tx names can be made like this:
h5file_link[["transcript_ids"]] <- names(tx) # GRangesList of all transcripts, must match gtf used by ribocode!