About the output files Salmon quant genes.sf

[ChristopheLegendre]

salmon.jst related information

The files with extension .genes.sf and filtered.genes.sf have only two columns : ENSG_ID and TPM
One is filtered the other is not.

Questions:

1. What is the type of filtering performed?
2. In addition to the filtering (row-wise), is there a recalculation of the TPM? if yes, can we add this information to the Phoenix Documentation?
3. Why the extension sf? I guess it is used by a specific downstream tool?

Suggestion:

1. Can we have the GeneName or GeneSymbol added to these files? These names are based on the Genomic Model so it should not be a problem to add the information
2. Can we have the data sorted based on the ENSG_ID?

Thanks.
PS: we may convert this discussion into an issue if needed.
C.

cat   MMRF_1157/rna/quant/salmon/MMRF_1157_1_BM_CD138pos_T2_TSMRU/MMRF_1157_1_BM_CD138pos_T2_TSMRU.genes.sf | grep ENSG00000110092    
 Name	Length	EffectiveLength	TPM	NumReads
ENSG00000110092	2726.31	2790.55	477.286	150254   

cat   MMRF_1157/rna/quant/salmon/MMRF_1157_1_BM_CD138pos_T2_TSMRU/MMRF_1157_1_BM_CD138pos_T2_TSMRU_filtered.genes.sf | grep ENSG00000110092  
Name  TPM
ENSG00000110092	1729.34  

In the filtered.genes.sf file, the TPM value is now 1729.34, which is quite different from the 477.286 in genes.sf file.

Which value should we use?

[ChristopheLegendre]

Answers to questions:

1. Some genes got removed and the TPMs were recalculated
2. Yes, some tools use that sf format which normally is expecting 5 columns as in the first example provided above
3. Data from the filtered file are more accurate regarding the TPMs values; TPMs were recalculated after having Igs genes (and b2m) removed

Answers to Suggestions

1. We could but it will/might break the expected sf format
2. I do not have the answer yet to that request

NOTE: either the filtered.genes.sf file mus be updated to have 5 columns as the unfiltered original file or the extension sf must be modified to match the relevant format of 2 columns (such as tsv)