salmon output interpretation

Question

[shinyjj]

Hi salmon users,

Can someone help me interpret salmon output? I ran salmon with one short read sample and see most of its TPM level is 0. I see 150653 among, 252045 is 0. Does this mean, for this specific sample, it's relative transcript abundance is very low or did I miss something? Is it normal to have the majority of the transcript to have low TPM level or is there a way to increase TPM?

[Shred]

Show us also the log file. Those quantification are relatives to transcripts: it seems legit.

[shinyjj]

hi shred, are you asking for salmon_quant.log file? There are other log files as well in index.

[shinyjj]

Here is the log file of the salmon_quant.log file

---

[2023-03-03 16:19:12.721] [jointLog] [info] setting maxHashResizeThreads to 8
[2023-03-03 16:19:12.721] [jointLog] [info] Fragment incompatibility prior below threshold.  Incompatible fragments will be ignored.
[2023-03-03 16:19:12.721] [jointLog] [info] Usage of --validateMappings implies use of minScoreFraction. Since not explicitly specified, it is being set to 0.65
[2023-03-03 16:19:12.721] [jointLog] [info] Setting consensusSlack to selective-alignment default of 0.35.
[2023-03-03 16:19:12.721] [jointLog] [info] parsing read library format
[2023-03-03 16:19:12.721] [jointLog] [info] There is 1 library.
[2023-03-03 16:19:12.722] [jointLog] [info] Loading pufferfish index
[2023-03-03 16:19:12.722] [jointLog] [info] Loading dense pufferfish index.
[2023-03-03 16:19:13.958] [jointLog] [info] done
[2023-03-03 16:19:13.999] [jointLog] [info] Index contained 252,045 targets
[2023-03-03 16:19:14.040] [jointLog] [info] Number of decoys : 0
[2023-03-03 16:19:14.425] [jointLog] [info] Automatically detected most likely library type as U

[2023-03-03 16:26:06.517] [jointLog] [info] Computed 756,724 rich equivalence classes for further processing
[2023-03-03 16:26:06.517] [jointLog] [info] Counted 29,086,612 total reads in the equivalence classes 
[2023-03-03 16:26:06.525] [jointLog] [info] Number of mappings discarded because of alignment score : 731,135,427
[2023-03-03 16:26:06.525] [jointLog] [info] Number of fragments entirely discarded because of alignment score : 10,648,839
[2023-03-03 16:26:06.525] [jointLog] [info] Number of fragments discarded because they are best-mapped to decoys : 0
[2023-03-03 16:26:06.525] [jointLog] [info] Number of fragments discarded because they have only dovetail (discordant) mappings to valid targets : 0
[2023-03-03 16:26:06.525] [jointLog] [info] Mapping rate = 63.5299%

[2023-03-03 16:26:06.525] [jointLog] [info] finished quantifyLibrary()
[2023-03-03 16:26:06.529] [jointLog] [info] Starting optimizer
[2023-03-03 16:26:06.678] [jointLog] [info] Marked 0 weighted equivalence classes as degenerate
[2023-03-03 16:26:06.698] [jointLog] [info] iteration = 0 | max rel diff. = 4821.77
[2023-03-03 16:26:06.518] [fileLog] [info] 
At end of round 0
==================
Observed 45784152 total fragments (45784152 in most recent round)

[2023-03-03 16:26:08.814] [jointLog] [info] iteration = 100 | max rel diff. = 17.9294
[2023-03-03 16:26:10.914] [jointLog] [info] iteration = 200 | max rel diff. = 20.672
[2023-03-03 16:26:13.017] [jointLog] [info] iteration = 300 | max rel diff. = 18.0116
[2023-03-03 16:26:15.122] [jointLog] [info] iteration = 400 | max rel diff. = 5.24671
[2023-03-03 16:26:17.208] [jointLog] [info] iteration = 500 | max rel diff. = 3.59377
[2023-03-03 16:26:19.343] [jointLog] [info] iteration = 600 | max rel diff. = 1.61581
[2023-03-03 16:26:21.409] [jointLog] [info] iteration = 700 | max rel diff. = 1.3212
[2023-03-03 16:26:23.514] [jointLog] [info] iteration = 800 | max rel diff. = 0.330426
[2023-03-03 16:26:25.626] [jointLog] [info] iteration = 900 | max rel diff. = 1.56766
[2023-03-03 16:26:27.714] [jointLog] [info] iteration = 1,000 | max rel diff. = 4.22352
[2023-03-03 16:26:29.842] [jointLog] [info] iteration = 1,100 | max rel diff. = 0.387334
[2023-03-03 16:26:31.953] [jointLog] [info] iteration = 1,200 | max rel diff. = 0.312137
[2023-03-03 16:26:34.079] [jointLog] [info] iteration = 1,300 | max rel diff. = 0.104816
[2023-03-03 16:26:36.192] [jointLog] [info] iteration = 1,400 | max rel diff. = 0.0577753
[2023-03-03 16:26:38.311] [jointLog] [info] iteration = 1,500 | max rel diff. = 0.0198889
[2023-03-03 16:26:40.427] [jointLog] [info] iteration = 1,600 | max rel diff. = 0.376883
[2023-03-03 16:26:42.449] [jointLog] [info] iteration = 1,696 | max rel diff. = 0.00400253
[2023-03-03 16:26:42.454] [jointLog] [info] Finished optimizer
[2023-03-03 16:26:42.454] [jointLog] [info] writing output 

Answer 1

[ATpoint]

Towards the TPM all that says is that about 60% of annotated transcripts have no counts, so likely not expressed. Is that unusual? I cannot say I ever felt the need to make such a plot, especially on transcript level but I think it's expected. No celltype expresses all annotated genes/transcripts as expression is highly specific in tissues and celltypes. It's probably fine. Just continue.

[Rob]

Yup; exactly as you say. One expects, in general, that most transcripts are not expressed in an individual sample. Many genes (and hence their isoforms) are unexpressed in individual samples. Further, for genes that are expressed, they often do not express all of their potential isoforms.

[shinyjj]

Thank you ATpoint and Rob :)