Coverage standards for RNA-sequencing

LP_SEP23 · 12-08-2014, 12:20 PM

Hi,

As part of my new faculty appointment I am the faculty adviser for my department helping get off the ground an Illumina sequencing core for the university.

While trying to put together some guidelines regarding sequencing coverage I became quite confused as to what is right,
Can anyone refer me to the most recent best practices or good papers dealing with this issue?

The original ENCODE recommendations do not agree much with my experience.
Outside the fact that you need at least 3 and not 2 biological replicates to do good stats the 30M PE reads do not seem enough according to my calculation bellow:

Given a Human Genome size of 3 billion bp, assuming that 80% of the reads will be mapped with high accuracy and estimating that 10% of the genome makes polyA RNA (this is the proportion of the genome I usually end up mapping to)
the average coverage of 30M 100 bp reads (0.03 billion reads) is: (0.03x100x0.8)/(3x0.1)= 8X
this seems really low, is my calculation correct?

is my mistake assuming that 10% of the genome gets mapped (if we assume 2% then you get 40X coverage, but that is not my experience)

thanks in advance for the feedback

HESmith · 12-08-2014, 12:34 PM

Coverage requirements depend upon your experiment. For differential gene expression with a well-annotated genome, biological triplicates at 15M single-end 50bp reads may suffice (e.g., see here). For isoform quantification or transcriptome assembly, 50M PE-100bp reads may be inadequate.

GenoMax · 12-08-2014, 12:47 PM

Depth recommendations from GenoHub (use as a guide).

https://genohub.com/next-generation-...g-guide/#depth

blancha · 12-08-2014, 01:12 PM

Average coverage is not meaningful for RNA-Seq.
Coverage is related to the level of expression of the gene.
More "reads will be captured from highly expressed genes, and few reads will be captured from genes expressed at low levels."

blancha · 12-08-2014, 05:49 PM

"A transcriptome represents that small percentage of the genetic code that is transcribed into RNA molecules — estimated to be less than 5% of the genome in humans (Frith et al., 2005)."
http://www.nature.com/scitable/topic...e-function-605

Genohub · 04-22-2015, 08:54 PM

Coverage and Read Depth by Sequencing Application - a new guide

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Coverage "standards" for SNP detection in tumor samples	giorgifm	Bioinformatics	7	06-26-2013 03:41 PM
RNASeq: Synthetic spike-in standards for RNA-seq experiments.	DZhang	Literature Watch	0	08-08-2011 07:09 AM
RNA-Seq: Synthetic spike-in standards for RNA-seq experiments.	Newsbot!	Literature Watch	0	08-06-2011 03:00 AM

12-08-2014, 12:20 PM	#1
LP_SEP23 Junior Member Location: Philadelphia Join Date: Feb 2012 Posts: 9	Coverage standards for RNA-sequencing Hi, As part of my new faculty appointment I am the faculty adviser for my department helping get off the ground an Illumina sequencing core for the university. While trying to put together some guidelines regarding sequencing coverage I became quite confused as to what is right, Can anyone refer me to the most recent best practices or good papers dealing with this issue? The original ENCODE recommendations do not agree much with my experience. Outside the fact that you need at least 3 and not 2 biological replicates to do good stats the 30M PE reads do not seem enough according to my calculation bellow: Given a Human Genome size of 3 billion bp, assuming that 80% of the reads will be mapped with high accuracy and estimating that 10% of the genome makes polyA RNA (this is the proportion of the genome I usually end up mapping to) the average coverage of 30M 100 bp reads (0.03 billion reads) is: (0.03x100x0.8)/(3x0.1)= 8X this seems really low, is my calculation correct? is my mistake assuming that 10% of the genome gets mapped (if we assume 2% then you get 40X coverage, but that is not my experience) thanks in advance for the feedback

12-08-2014, 12:34 PM	#2
HESmith Senior Member Location: Bethesda MD Join Date: Oct 2009 Posts: 509	Coverage requirements depend upon your experiment. For differential gene expression with a well-annotated genome, biological triplicates at 15M single-end 50bp reads may suffice (e.g., see here). For isoform quantification or transcriptome assembly, 50M PE-100bp reads may be inadequate.

12-08-2014, 12:47 PM	#3
GenoMax Senior Member Location: East Coast USA Join Date: Feb 2008 Posts: 7,088	Depth recommendations from GenoHub (use as a guide). https://genohub.com/next-generation-...g-guide/#depth

12-08-2014, 01:12 PM	#4
blancha Senior Member Location: Montreal Join Date: May 2013 Posts: 367	Average coverage is not meaningful for RNA-Seq. Coverage is related to the level of expression of the gene. More "reads will be captured from highly expressed genes, and few reads will be captured from genes expressed at low levels."

12-08-2014, 05:49 PM	#5
blancha Senior Member Location: Montreal Join Date: May 2013 Posts: 367	"A transcriptome represents that small percentage of the genetic code that is transcribed into RNA molecules — estimated to be less than 5% of the genome in humans (Frith et al., 2005)." http://www.nature.com/scitable/topic...e-function-605

04-22-2015, 08:54 PM	#6
Genohub Registered Vendor Location: genohub.com Join Date: Mar 2013 Posts: 210	Coverage and Read Depth by Sequencing Application - a new guide