I am new in the field on RNA-Seq analysis and trying to understand the methodology behind Salmon's aggregation of transcripts into gene level.
From the forum https://github.com/COMBINE-lab/salmon/issues/98 The answer is "Salmon's aggregation strategy is to collect the transcript-level abundances and then aggregate them to the gene level (this means summing the counts / TPMs, and doing an abundance weighted combination of the lengths / effective lengths)." However, I was not able to understand is properly.
can someone explain it in detail for this example where the quantification at transcript level and its aggregated output at gene level are shown?
Name Length EffectiveLength TPM NumReads
ENSG00000172530 78.6693 60.7441 9.11087 516.593
ENST00000569400 802 553 0.0341855 1.27033
ENST00000423252 644 395 0.0573103 1.52117
ENST00000488074 759 510 0 0
ENST00000439677 736 487 1.10087 36.0257
ENST00000454563 558 309 0 0
ENST00000479780 1551 1302 0.382663 33.4793
ENST00000393208 2283 2034 0.374088 51.1297
ENST00000459966 627 378 0 0
ENST00000412691 586 337 1.58787 35.9578
ENST00000526460 547 298 0 0
Moreover, I will like to know how it is different from simplesum_avetxl used in Tximport mentioned in the paper "Differential analyses for RNA-seq: transcript-level estimates improve gene-level inferences"?
In addition, will it be wrong to use "quant.genes.sf" instead of "quants.sf" as input for Tximport? if yes, then why?
From the forum https://github.com/COMBINE-lab/salmon/issues/98 The answer is "Salmon's aggregation strategy is to collect the transcript-level abundances and then aggregate them to the gene level (this means summing the counts / TPMs, and doing an abundance weighted combination of the lengths / effective lengths)." However, I was not able to understand is properly.
can someone explain it in detail for this example where the quantification at transcript level and its aggregated output at gene level are shown?
Name Length EffectiveLength TPM NumReads
ENSG00000172530 78.6693 60.7441 9.11087 516.593
ENST00000569400 802 553 0.0341855 1.27033
ENST00000423252 644 395 0.0573103 1.52117
ENST00000488074 759 510 0 0
ENST00000439677 736 487 1.10087 36.0257
ENST00000454563 558 309 0 0
ENST00000479780 1551 1302 0.382663 33.4793
ENST00000393208 2283 2034 0.374088 51.1297
ENST00000459966 627 378 0 0
ENST00000412691 586 337 1.58787 35.9578
ENST00000526460 547 298 0 0
Moreover, I will like to know how it is different from simplesum_avetxl used in Tximport mentioned in the paper "Differential analyses for RNA-seq: transcript-level estimates improve gene-level inferences"?
In addition, will it be wrong to use "quant.genes.sf" instead of "quants.sf" as input for Tximport? if yes, then why?