Hi PRinlgler,
did you come up with a solution?
I'm also quite new and tried to understand what is the mathematical process from cufflinks to cuffdiff...why are the fpkm in cufflinks differnt from those given in cuffdiff??
thanks,
ib
Seqanswers Leaderboard Ad
Collapse
Announcement
Collapse
No announcement yet.
X

Cuffdiff FPKM and test statistic calculations
Hello, I have been lurking here a bit, but haven't posted yet, so bear with me.
I am trying to recreate the calculations of FPKM and the test statistic performed by uffdiff on two sample genes in order to better understand how my RNAseq data has been tested (and thus better explain my data). I have been reading through the supplemental information from the paper linked on the Cufflinks website (Transcript assembly and quantification by RNASeq reveals unannotated transcripts and isoform switching during cell differentiation; Nature Biotechnology 28,511–515(2010)doi:10.1038/nbt.1621), and I have been able to get through part of the calculations, but am still stumped.
To calculate FPKM, I am using the following formula:
(10^9)*Xg*(gamma t)
~l(t)*M
Where:
Xg is the number of fragments aligned to the gene locus (g),
gamma t is the maximum likelihood estimate of the probability of selecting a fragment from a transcript (t) coming from that gene locus
~l(t) is the adjusted transcript length
Σ from i=1 to t(i)) [F(i)(l(t)i+1)]
F(i) = probability that the fragment has a length i
l(t) = the transcript length
M is the total number of fragments mapped in that sample
My sample is run with singleend reads 50 bp long, so I simplified the ~l(t) to the transcript length  49, and we ran cuffdiff using a GTF file that contains only the loci for one transcript for each gene, so I think gamma should be 1. One of my lab members wrote a program that assigns raw reads to individual genes, so I am using the reads from this program for the raw reads in the FPKM calculation. Using this data I get pretty close to the FPKM calculated:
Code:gene Reads A Reads B Cuff A Cuff B Calc A Calc B A 2 81 0.021956 0.823028 0.02116 0.795515 B 1 40 0.007541 0.279201 0.007374 0.273778
From this, I get into much more complicated math when trying to recreate the test statistic. The formula for the test statistic, from the supplemental data, is:
[log(Xa)+log(gamma a)+ log(Mb)  log(Xb)  log(gamma b)  log (Ma)]
SQRT[ (psi a*(1+Xa)*(gamma a)^2)/(Xa*(gamma a)^2)+ (psi b*(1+Xb)*(gamma b)^2)/(Xb*(gamma b)^2)]
Where psi is a variancecovariance matrix that estimates the covariance between gamma (tk) and gamma (tl). As far as I can tell, tk and tl come from Tophat, which splits a read of length l into two smaller reads of length k in order to align across splice junctions. (This may be completely off base)
When I run the equations using these assumptions, though, I get a value far off from the test statistic calculated by Cuffdiff:
Code:Gene T stat Cufflinks tstat A 0.99128963 3.42295 B 0.898803344 3.07139
So, I suppose my question is, is there a way for me to calculate (or even estimate) the psi function? Am I making faulty assumptions?
I apologize if this question has been addressed in a previous thread. I did try to find the answer in the archives before asking here.
Thank youTags: None
Latest Articles
Collapse

by seqadmin
Proteins are often described as the workhorses of the cell, and identifying their sequences is key to understanding their role in biological processes and disease. Currently, the most common technique used to determine protein sequences is mass spectrometry. While still a valuable tool, mass spectrometry faces several limitations and requires a highly experienced scientist familiar with the equipment to operate it. Additionally, other proteomic methods, like affinity assays, are constrained...
Channel: Articles
04042024, 04:25 PM 

by seqadmin
Despite advancements in sequencing platforms and related sample preparation technologies, certain sample types continue to present significant challenges that can compromise sequencing results. Pedro Echave, Senior Manager of the Global Business Segment at Revvity, explained that the success of a sequencing experiment ultimately depends on the amount and integrity of the nucleic acid template (RNA or DNA) obtained from a sample. “The better the quality of the nucleic acid isolated...
Channel: Articles
03222024, 06:39 AM 
ad_right_rmr
Collapse
News
Collapse
Topics  Statistics  Last Post  

Started by seqadmin, Yesterday, 12:08 PM

0 responses
11 views
0 likes

Last Post
by seqadmin
Yesterday, 12:08 PM


Started by seqadmin, 04102024, 10:19 PM

0 responses
14 views
0 likes

Last Post
by seqadmin
04102024, 10:19 PM


Started by seqadmin, 04102024, 09:21 AM

0 responses
14 views
0 likes

Last Post
by seqadmin
04102024, 09:21 AM


Started by seqadmin, 04042024, 09:00 AM

0 responses
43 views
0 likes

Last Post
by seqadmin
04042024, 09:00 AM

Leave a comment: