Different results for the same data using Trinity. Why?

Question

[tawares07]

Hi guys!

I used Trinity version (r20140717) twice for the same data (run SRR1462347) and I have got a small difference in the number of assembled transcripts (Trinity.fasta).

How this difference can be explained? Is it acceptable?

Thank you very much!

[st.ph.n]

Please post the commands/parameters you used for both runs.

[tawares07]

Command:

.../TRINITY/trinityrnaseq_r20140717/Trinity --no_bowtie --seqType fq --JM 40G --left [file_1] --right [file_2] --CPU 10 --output [dir_out]

thanks!

[SES]

This may be due to using multiple processors. Try using just 1 CPU and I would bet the results are the same. I'm not certain, but this is true for Velvet anyway (i.e., different results with multiple CPUs, same result with 1 CPU).

[tawares07]

Hi, SES!

I did what you suggested. I will share the results as soon as possible.

Thanks =)

Answer 1

[rick.westerman]

Trinity is non-deterministic. I believe this arises because of figuring out alternative transcripts will depend on which base transcript you start off with and the starting transcript is random. Thus small changes are acceptable.