[Denovoassembler-devel] RE : Why flush both k-mer and its complement?

[Sébastien B. <seb...@ul...>]

I think there is a bug:

In your for loop, the second iteration (flush=1) should add b in the m_bufferedData, not a twice !



What I had in mind (pseudo C++):

/* get the pair of k-mers */
Kmer a = kmerAtPosition()
Kmer b=a.reverseComplement()

/* compute the destination */
Kmer rankToFlush=-1;

if(a<b)
	rankToFlush=a.hash_function_1()%m_parameters.getSize()
else
	rankToFlush=b.hash_function_1()%m_parameters.getSize()


/* push a to destination and flush if necessary */
push a to rankToFlush usin bufferedData.addAt
bufferedData.flush()

/* push b to destination and flush if necessary */
push b to rankToFlush usins bufferedData.addAt
bufferedData.flush()


This has only a call to reverseComplement and do the same as refs/master


> ________________________________________
> De : David Eccles (gringer) [dav...@mp...]
> Date d'envoi : 21 juillet 2011 11:07
> À : Sébastien Boisvert
> Cc : den...@li...
> Objet : Re: Why flush both k-mer and its complement?
> 
> On 21/07/11 16:40, Sébastien Boisvert wrote:
>> Although both k-mers will go on the same MPI rank,
>> the first flush may generate an actual MPI message on the network.
>> As such, the second flush won't flush anything because the buffer was just reset.
>> On the other hand, if the first k-mer does not generate a message when flush is called,
>> then the second flush may do so.
>>
>> Feel free to send me a patch for this !
> 
> Okay, I've attached a patch that just does the flush twice without any
> additional reverse complelements.
> 
> -- David
> 

                                                     Sébastien