Resolving Salmon Errors?

Question

[pearl2070]

I'm running Salmon with this line:

*./salmon-1.8.0_linux_x86_64/bin/salmon quant -p 12 -t microbial_all_cds.fasta -l A -a Sample1_megahit.annotation_bwa.sam -o Sample1_salmon*

And I encounter these errors:

*" Transcript appears twice in the transcript FASTA file" and "Transcript appears in the reference but did not appear in the BAM."*

Is there a way to, in command line,

1. process my transcript FASTA file so that there are no duplicates and
2. process my reference file so that it's filtered, containing only transcripts that are also in my SAM file?

[rpolicastro]

When you generated your BAM did you align your reads to the genome or transcriptome? Salmon requires alignment to the transcriptome. You may also want to include your code for that part.

[pearl2070]

Ah, I aligned the reads to a reference database (microbial_all_cds.fasta). For BWA, I ran:

*bwa index -a bwtsw microbial_all_cds.fasta*

Followed by:

*bwa mem -t 32 microbial_all_cds.fasta Sample1_megahit.contigs.fa > Sample1_megahit.annotation_bwa.sam*

Should I instead be doing:

*bwa index -a bwtsw Sample1_megahit.contigs.fa*

*bwa mem -t 32 Sample1_megahit.contigs.fa Sample1__mRNA.fa > Sample1_megahit.annotation_bwa.sam*

And then:

*./salmon-1.8.0_linux_x86_64/bin/salmon quant -p 12 -t Sample1_megahit.contigs.fa -l A -a Sample1_megahit.annotation_bwa.sam -o Sample1_salmon*

[GenoMax]

If Sample1_megahit.contigs.fa is your assembled transcriptome then yes.

[pearl2070]

Got it, thank you!

[GenoMax]

Prior reference: Running salmon in alignment mode?