genes.fpkm_tracking (Coverage column)

[angerusso]

In the genes.fpkm_tracking file, why is the "coverage" for the gene not reported from cufflinks?

I understand to get the fpkm_per_gene, transcript level fpkm is summed up. How about coverage column for gene. Does this get summed up from transcript level as well or in other words, why is column reported if not computed?

tracking_id class_code nearest_ref_id gene_id gene_short_name tss_id locus length coverage FPKM FPKM_conf_lo FPKM_conf_hi FPKM_status
OR4F3 - - OR4F3 OR4F3 TSS4836 chr1:367658-368597 - - 0 0 0 OK
FAM138A - - FAM138A FAM138A TSS8229 chr1:34610-36081 - - 0 0 0 OK
OR4F5 - - OR4F5 OR4F5 TSS14112 chr1:69090-70008 - - 0 0 0 OK
OR4F16 - - OR4F16 OR4F16 TSS12406 chr1:621095-622034 - - 0 0 0 OK
LOC100288069 - - LOC100288069 LOC100288069 TSS8527 chr1:700244-714068 - - 1.1941 0 2.26074 OK
LOC643837 - - LOC643837 LOC643837 TSS10020 chr1:763063-789740 - - 1.42148 0 2.81223 OK
LINC00115 - - LINC00115 LINC00115 TSS17932 chr1:761585-762902 - - 0.57203 0 1.41206 OK

[rflrob]

Does this get summed up from transcript level as well or in other words, why is column reported if not computed?

It's almost certainly that way so that if you write something to deal with a fpkm_tracking file, you'll always know that it has the same number of columns. The generic format is described here: http://cufflinks.cbcb.umd.edu/manual...racking_format. If tools that relied on that format had to check whether or not an uncomputed column was present, it would be a lot more complicated.