I am looking for the best approach to assigning read counts to de novo assembly contigs in order to analyze expression levels via RPKM. In doing this work I have thus far followed the Simple Fool's Guide gene expression analysis tutorial but this entails an approach where only reads that map uniquely are counted. My understanding is that some programs will assign reads based on an educated guess as to where a read may be assigned if it maps to multiple contigs -- and that these guesses may only be marginally better than a random assignment.
So my question is, what tools do people in this forum favor for assigning read counts to contigs for a de novo transcriptome assembly? Has unique read mapping been superseded by newer approaches? Also, my best mapping results have been generated with BWA-mem (approximately 83% paired end read mapping) so I am inclined to use a tool like DEseq, but at this point do not fully understand how it assigns reads.
Thanks in advance for any input.