using RSEM with non Trinity assembly

Question

[jen]

Hi all,

I am trying to use RSEM to receive relative abundance estimates of viruses within my metagenomic data, not transcripts. The problem is that I am no longer using Trinity as the assembler because I had much better luck with SPAdes, especially considering that I am not working with mRNA sequencing. Is it possible to use RSEM on a non Trinity assembly? If so, how do I go about doing so?

Note: I know that the SPAdes IDs provide a coverage meaurement for each individual contig; however, I need to compare the abundances of sequences across different samples and RSEM was good for that. What I mean is that I was able to pool all of my samples into one assembly, and RSEM would be able to differentiate the abundance of a certain sequence in each individual sample. If a sequence was assembled but only in one of my samples, RSEM would give me that information with the TPM measurements. If there is an alternative program I can use, please let me know!

[Dunois]

RSEM and Trinity are completely independent of one another. I don't see what the problem here actually is.

[jen]

Apologies for the confusion. I've successfully assembled my RNA data using SPAdes (I've had more success with this assembler for viral RNA work). I am wondering if I can perform the same tasks below with SPAdes assembly output rather than Trinity.

Here is my code for the Trinity assembly. The samples file contains forward and reverse reads for three different samples (six files total):

nohup /usr/local/bin/trinityrnaseq-v2.13.2/Trinity --seqType fq --max_memory 50G --samples_file  /data/users/jherrera/FL_LA_MD_samplesfile.txt --CPU 6 &

Prepping a reference for RSEM:

usr/local/bin/trinityrnaseq-v2.10.0/util/align_and_estimate_abundance.pl --transcripts trinity_out_dir.Trinity.fasta --prep_reference --trinity_mode --est_method RSEM --aln_method bowtie2

Then, I align reads to the assembly using bowtie2 and estimate abundance:

nohup  /usr/local/bin/trinityrnaseq-v2.10.0/util/align_and_estimate_abundance.pl --transcripts trinity_out_dir.Trinity.fasta --seqType fq --samples_file  /data/users/jherrera/FL_LA_MD_samplesfile.txt --est_method RSEM --aln_method bowtie2 --trinity_mode --output_dir RSEMoutput  &

The resulting RSEM.isoforms.results file produced for each of my samples provides me with TPM values for each contig that was assembled:

Essentially, I want to know if there's a way to carry out this same task (calculating TPMs per sample for each contig in my assembly) without using any scripts that are specific to Trinity. I am not well-experienced with any of this, so please let me know if that makes sense. I can provide more information if necessary. I've tried several things over the past few months but to no avail (I stepped away from this analysis for a while).

Thanks!

Answer 1

[Dunois]

Essentially, I want to know if there's a way to carry out this same task (calculating TPMs per sample for each contig in my assembly) without using any scripts that are specific to Trinity.

The short answer is, "Yes".

You just need to call RSEM directly and make it do what these perl scripts did here. I would actually recommend that you use salmon or kallisto instead. It'll be a one-step process with one of those and the abundance estimation will be much faster (while being nearly as accurate).

[jen]

Thanks! I'll try it out.