Tophat/cufflinks workflow question

[staylor]

If I have 6 samples, call them 1,2,3,4,5,6 (they could be 6 lanes or a 6 way multiplexed experiement). Then I want to compare 1,2,3 vs 4,5,6.

So if I map the reads using tophat, I assume I can run all the 1,2,3,4,5,6 in parallel and then cat and sort the relavant sam files together before I run cufflinks.

So

cat tophat1 tophat2 tophat3 | sort > 123.sam
cat tophat4 tophat5 tophat6 | sort > 456.sam

Then if I run cufflinks on 123.sam and 456.sam I get 123.transcripts.gtf and 456.transcripts.gtf.

So I was wondering what is the next step? Can I do

cuffcompare 123.transcripts.gtf -r geneannotations.gtf
cuffcompare 456.transcripts.gtf -r geneannotations.gtf

(geneannotations.gtf is an Ensembl annotations file)

so I get

123.transcripts.tmap and 456.transcripts.tmap

then I just compare the two data sets RPKM values using Ensembl ID as the in the same way you would use probe_id in a traditional array experiement?

Do I need to do any further normalization?

Am I missing something!?

Thanks in advance!

[Cole Trapnell]

Then if I run cufflinks on 123.sam and 456.sam I get 123.transcripts.gtf and 456.transcripts.gtf.

So I was wondering what is the next step? Can I do

cuffcompare 123.transcripts.gtf -r geneannotations.gtf
cuffcompare 456.transcripts.gtf -r geneannotations.gtf

(geneannotations.gtf is an Ensembl annotations file)

You probably want:
cuffcompare -r geneannotations.gtf 123.transcripts.gtf

Looking then at the out.tracking file will then tell you how the Ensembl transcripts (along with whatever new transcripts you find) compare to each other in the two samples.

Do I need to do any further normalization?

That's a tougher question. RPKM is a measure of relative gene expression designed to allow comparisons between experiments, just as you are asking to do. There are a fair number of people working on ways of correcting and normalizing for various biases in RNA-Seq, but there are no widely accepted means of doing so that I am aware of.

[staylor]

You probably want:
cuffcompare -r geneannotations.gtf 123.transcripts.gtf

Looking then at the out.tracking file will then tell you how the Ensembl transcripts (along with whatever new transcripts you find) compare to each other in the two samples.

Do you mean I need to do

cuffcompare -r geneannotations.gtf 123.transcripts.gtf 456.transcripts.gtf

otherwise how will the tracking file know about the two samples?

That's a tougher question. RPKM is a measure of relative gene expression designed to allow comparisons between experiments, just as you are asking to do. There are a fair number of people working on ways of correcting and normalizing for various biases in RNA-Seq, but there are no widely accepted means of doing so that I am aware of.

Ok. Thanks for your help on this.

[Cole Trapnell]

Do you mean I need to do

cuffcompare -r geneannotations.gtf 123.transcripts.gtf 456.transcripts.gtf

otherwise how will the tracking file know about the two samples?

Yes, sorry. That's what I get for responding before coffee.

[staylor]

:-) No problem.

Thanks for your help.

[staylor]

I did

cuffcompare -r geneannotations.gtf 123.transcripts.gtf 456.transcripts.gtf

and got:

- - q1:CUFF.49|CUFF.49.0|100|320.741503|0.000000|0.000000|uniq -
- - q1:CUFF.922|CUFF.922.0|100|2080.321131|0.211081|0.210872|uniq -
- - q1:CUFF.2093|CUFF.2093.0|100|320.213468|0.000000|0.000000|uniq -
- - q1:CUFF.923|CUFF.923.1|100|2498.467902|0.159820|0.159979|uniq -
- - q1:CUFF.2964|CUFF.2964.0|100|136.004645|0.000000|0.000000|uniq -
- - q1:CUFF.3939|CUFF.3939.0|100|2146.036538|0.000000|0.000000|uniq -
- - q1:CUFF.4125|CUFF.4125.0|100|612.020902|0.000000|0.000000|uniq -
- - q1:CUFF.4410|CUFF.4410.0|100|4534.343539|0.000000|0.000000|uniq -
- - q1:CUFF.5384|CUFF.5384.0|100|1515.019712|0.000000|0.000000|uniq -
- - q1:CUFF.5667|CUFF.5667.0|100|1515.969648|0.000000|0.000000|uniq -
- - q1:CUFF.5952|CUFF.5952.0|100|5568.492065|0.000000|0.000000|uniq -
- - q1:CUFF.6333|CUFF.6333.0|100|2299.714905|0.000000|0.000000|uniq -
- - - q2:CUFF.611|CUFF.611.0|100|2209.360572|0.000000|0.000000|uniq
- - q1:CUFF.7536|CUFF.7536.0|100|702.690665|0.000000|0.000000|uniq -
- - q1:CUFF.7629|CUFF.7629.0|100|3756.201011|0.000000|0.000000|uniq -
- - q1:CUFF.9212|CUFF.9212.0|100|417.816071|0.000000|0.000000|uniq -
- - q1:CUFF.9554|CUFF.9554.0|100|9625.347396|0.000000|0.000000|uniq -
- - q1:CUFF.9797|CUFF.9797.0|100|2347.966478|0.000000|0.000000|uniq -
- - q1:CUFF.9965|CUFF.9965.0|100|702.690665|0.000000|0.000000|uniq -
- - q1:CUFF.8952|CUFF.8952.0|100|719.829462|0.000000|0.000000|uniq -
- - q1:CUFF.10691|CUFF.10691.0|100|1806.918854|0.000000|0.000000|uniq -
- - q1:CUFF.10799|CUFF.10799.1|100|3619.236510|0.284840|0.284870|uniq -
- - q1:CUFF.10839|CUFF.10839.0|100|2153.582326|0.000000|0.000000|uniq -
- - q1:CUFF.11538|CUFF.11538.0|100|7683.301808|0.000000|0.000000|uniq -
- - q1:CUFF.10782|CUFF.10782.0|100|89.705191|0.000000|0.000000|uniq -
- - q1:CUFF.10800|CUFF.10800.0|100|3902.596250|0.174811|0.174792|uniq -
- - q1:CUFF.10825|CUFF.10825.0|100|6510.429805|0.000000|0.000000|uniq -
- - q1:CUFF.12645|CUFF.12645.0|100|58.557555|0.000000|0.000000|uniq -
- - q1:CUFF.13320|CUFF.13320.0|100|1702.673535|0.000000|0.000000|uniq -
- - q1:CUFF.12088|CUFF.12088.0|100|6322.264673|0.000000|0.000000|uniq -
- - q1:CUFF.13147|CUFF.13147.0|100|8946.451886|0.000000|0.000000|uniq -
- - q1:CUFF.14256|CUFF.14256.0|100|782.998170|0.000000|0.000000|uniq -
- - q1:CUFF.14910|CUFF.14910.0|100|1953.822826|0.000000|0.000000|uniq -
- - - q2:CUFF.1367|CUFF.1367.0|100|2051.549103|0.000000|0.000000|uniq
- - q1:CUFF.15282|CUFF.15282.0|100|1686.457597|0.000000|0.000000|uniq -
- - q1:CUFF.15535|CUFF.15535.0|100|2326.148410|0.000000|0.000000|uniq -
- - q1:CUFF.16741|CUFF.16741.0|100|1297.275075|0.000000|0.000000|uniq -
- - q1:CUFF.17421|CUFF.17421.0|100|3055.953562|0.000000|0.000000|uniq -
- - q1:CUFF.17710|CUFF.17710.0|100|2845.146363|0.000000|0.000000|uniq -
- - q1:CUFF.19743|CUFF.19743.0|100|133.631524|0.000000|0.000000|uniq -
- - q1:CUFF.19375|CUFF.19375.0|100|2369.916793|0.000000|0.000000|uniq -
- - q1:CUFF.19582|CUFF.19582.0|100|4355.505260|0.000000|0.000000|uniq -
- - q1:CUFF.19789|CUFF.19789.0|100|5320.372181|0.000000|0.000000|uniq -
- - q1:CUFF.22524|CUFF.22524.0|100|647.153215|0.000000|0.000000|uniq -
- - q1:CUFF.22483|CUFF.22483.0|100|21957.950620|0.000000|0.000000|uniq -
- - q1:CUFF.22641|CUFF.22641.0|100|13234.040114|0.000000|0.000000|uniq -

Does this mean there are no overlaps for annotated genes? This sample was mutiplexed on 1 lane so there aren't as many reads as you would normally expect. Would this have something to do with it?

[townway]

In cufflinks document, cufflinks only accepted input SAM format by “sort –k 3,3 –k 4,4n *.sam”. but when I use the *sam format generated by BWA, and type sort command, the format change from

NCI-GA2:1:1:4:358#0 147 chr1 53372939 60 35M = 53372816 -158 ATGGGCTGGAT
GATCCCTGTTCAGGCCTAATCCGC A>BAA>=BBBBBB@>?BBBA@8>BB>BBABB>BCB XT:A:U NM:i:0 SM:i:37 AM:i:23 X0:i:1 X1:
i:0 XM:i:0 XO:i:0 XG:i:0 MD:Z:35
NCI-GA2:1:1:4:1683#0 99 chrUn_gl000220 159630 0 35M = 159744 149 CTAGGGCGCGGGCCCGGGT
GGAGCCGCCGCAGGTG BB@A=?B<B?:>BBBAA<9BABABBBBB<BA=78A XT:A:R NM:i:0 SM:i:0 AM:i:0 X0:i:2 X1:i:0 XM:
i:0 XO:i:0 XG:i:0 MD:Z:35

to

NCI-GA2:1:100:1000:1645#0 133 * 0 0 * * 0 0 TCCTCCTTTTTCACTTGAT
CCCACCGATGTCTGCC BCBCCCBBCCCBCB@BACABBCCCA>BAAABABBB
NCI-GA2:1:100:1000:1645#0 69 * 0 0 * * 0 0 CAAGTCTGCATGGCTGTTG
ACATAGGCAGACATCG BCCB=ABBABB@@AA@;>;A75/86=376;:;/9:
NCI-GA2:1:100:1000:391#0 133 * 0 0 * * 0 0 TACCGCGGCTGCTGGCACC
AGACTTGCCCAGATCG BAAAAABBAB>5>7998;<1=9:A@@>A@=>;=7=

Some part of the line are missing, so cufflinks would not recognize it. How can I change the sort parameter to get correct format for cufflinks?

[staylor]

Try filtering for hits only (simple way is grep for chr if your genome uses this in the fasta header line) and then sort.