PureBasic Forums - English

@Wilbert Even if it is more convenient, the focus is on controllable memory- and time consumtion on desktop See, k-mering is to "find the unknown", may be the high counts or the frequencies in a longer genome, or an unknown long sequence. The sorting & grouping can though be a selectab...

@Wilbert yes, you did. With canonical reverse it's approx. 20% slower. Conceptional code: RComplement = DNAsearchSeq For z = 0 To KmerLen - 1 ; Ascii PB5.40 ; XLAT single-byte-instruction, mit 7/1c oder 7/2 with small table in DataSection H_DIGIT = PeekA(@RComplement + z) - 65 MOV AL, [v_H_DIGIT] ; ...

@Wilbert

As a fallback I tried yor version with k=31-31 for step by step K-mer, but it didn't work?

QWilbert As I descibed, it's the fact finding stage. "The more and the quicker" is the goal, because a next generation illumina sequencer is able to produce more than 2 terabases of data per week (Illumina sequencing is a analog process) Up to now I don't know how many machines are working...

@Wilbert Unfortunately the reply function swallowed my last reply-draft. So once again: Ad hoc I can confirm the memory requirements of my more or less "conventional" PB-map/array approach for ecoli 1-12. In any way I don't want to disturb your creative phase for a new idea, but I want to ...

@Wilbert Sure, you are right, there should be a goal. Of course one could define this as "Count & qualify uniques in k-mer up to 1-31 of the human genome presented in state of the art graphics in an desktop environment under (small & fast) PB-Windows". My inspiration was the backlo...

That happens with quick and dirty: "DNAsearchSeq.s" in the structure is not used, it is extracted from the ReadData-Buffer with DNAindex and k-mer length. At that point I saw your 2015-ASM and contacted you.

@Wilbert Right now the timing with phi-X174 is not the most important, because it's small and 14 ms vs. 43 ms is close together. More meaningful is the timing with ecoli which is a thousandfold larger. Here I hoped a lot to your new approch, but the result is striking: As I described the 1-12 ecoli ...

@Wilbert,
I may be this is of interest:
phi-X174 produces a frequency of 34.181
ecoli produces a frequency of 6.917.666

@Wilbert, Test results: - No errors observed - The results of 5.386 bp phi-X174 genome with your pgm are the same as with mine pgm - My pgm needs 43 ms to produce the results on phi-X174, your pgm needs 77 ms - Memory demands: on phi-X174 about the same Your questions: - 1-1-Frequency: 4, file-size/...

@Idle,
what happened to your test?
Do you need more input?

I am eager to your observations and hope your system didn't go to smoke...

@Wilbert, first of all, I like your clear coding style. We can't and don't need to carry the positions of occurence with it. Today I use the position to recover the string in combination of the k-mer length during sorting and file-output. So, for your approch, we don't need them. Visible uniqueness:...

@Wilbert, as a standard, I use 64-bit executables. I jumped into the topic since the time that the output-capacities of DNA sequencers grew faster than the computing capacities (A single next-generation sequencer easily generates billions of short DNA sequences (approx. length 200 each). This outrac...

@Idle: The performance is pretty dismal, though that's not surprising given that its not threaded I ment it hopeful: 250 KB DNA-Input (the small phi-X174 has only 5.386 bp), the k's from 1 to 31 evaluated, 260 mb memory in 30 secs.... First hand not too bad. I could estimate your idea better, if you...

@Idle, thank you very much for the conclusions in a trie, even if it does'nt help the k-problem. "... telling you an item exists when it doesn't." --> My spontaneos thought "OK, we have the potential needles now, let's verify them with something like PCMPEQB or PCMPISTRI" was van...