search

blahah / transrate

Understand your transcriptome assembly
http://hibberdlab.com/transrate
Other
100 stars 34 forks source link

Snap failing due to memory requirements on large transcriptomes #225

Open kvpmulder opened 5 years ago

kvpmulder commented 5 years ago

Hi transrate community,

I have a trinity assembly and want to run a total of 4 HiSeq lanes of data against it with transrate. For snap to run successfully I need to increase the mcp threshold and the memory usage becomes too high. I can only run about 1/3 of the data in one go without hitting the limit for our server (450Gb). This seems very high for a mapping step. Is there a way to reduce the memory usage of snap, or a good way to run the data in subsets and then combine the results for several transrate runs done on the same assembly?

Thank you!

Kevin

MDSharma commented 5 years ago

Hi, any thoughts?

kvpmulder commented 5 years ago

i ended up doing an in silico normalization using trinity and then only using the normalized reads to run transrate. Its not ideal because you lose some power but it reduces the memory threshold needed.