Re: [Cortexassembler-variation] incompatible error

[Zamin I. <za...@we...>]

Hi Andrew

This is just a confusion about single and multicolour graphs. I'll explain:


>    I have sequencing data of 203 samples. I builded uncleaned individual
>graph for each sample and then merged all of the individual binaries into a
>pooled graph with the parameter "--remove_low_coverage_supernodes 1". I
>want to clean each individual by comparing it with the cleaned pool graph.


If you load fasta, cortex always dumps a single-colour graph, independent of how many colours
you have compiled for.

If you load binaries, then when you use --dump_binary,  cortex dumps a graph with the same number of colours as you have compiled for.

So, you correctly built single-colour binaries for your 203 samples.
Then you wanted to pool them, so that you could clean the pool. The right thing to do was to use a ONE-COLOUR graph,
and load them all into one colour, and then clean that single colour graph.  Then you could load that 1-colour graph using 
Cortex compiled for 2-colours, leaving one colour spare for loading and cleaning each sample one at a time.

Unfortunately, what you did was to use a 201-colour graph and load each sample into a different colour. 
You then cleaned that and dumped a 201-colour graph.
Then you tried reloading that 201-colour graph using Cortex compiled for 2 colours

That's why Cortex complained and said
"You are loading  binary with 203 colours into a graph with 2 colours - incompatible".

All you need to do is make a single list of all your sample binaries, and load that into a single-colour
COrtex, and clean them again, and dump a 1-colour binary. Then try you final cleaning commandline again

regards

Zam








		

------------------------------------------------------
Zamin Iqbal
Postdoctoral Researcher
Wellcome Trust Centre for Human Genetics
University of Oxford

www.well.ox.ac.uk/~zam