Question about ONT's Developer Agreement

[ymc]

Anyone here signed ONT's Developer Agreement?

I read the fine terms. It seems like anything that built on top of their developer kit will become ONT's intellectual property. Is that true?

For example, Matt Loose et al's selective sequencing paper:

http://biorxiv.org/content/early/2016/02/03/038760

uses the ReadUntil capability which is part of the developer kit. Does that mean everything Loose et al developed in that preprint belongs to ONT???

[gringer]

I've refused to sign ONT's developer agreement because of the restrictions of use that are claimed by ONT. I much prefer waiting a couple more months and using free software.

[ymc]

oic. No wonder there were no follow-up to that Matt Loose paper.

[gringer]

The "read until" is still being developed, but Matt doesn't have much time to specifically work on it, and it's mostly useless now with a sequencing speed of 450 bases per second. Computing matches just takes far too long -- by the time a program has decided whether or not to keep a sequence, a substantial fraction of the reads will have already passed through the pore.

ONT are currently developing FPGA methods that can make read-until decisions at millisecond timescales, which would bring read-until back into the realm of possibility. A really nice application of read-until is for direct RNA sequencing from total RNA. Imagine being able to decide within 1/10 of a second (or 50bp sequenced) whether or not a sequence is ribosomal RNA, and ejecting anything that fits that profile, freeing up sequencing capacity for non-ribosomal sequences: no more need for ribosomal depletion.

[ymc]

Interesting, I am also doing something along the line. Nevertheless, I am doing 16S DNA not RNA but presumably should be applicable to RNA. I have a hack specific to 16S that allows me to align R9 data with DTW in O(m*n) time instead of O(m*n*o) where m is 16S length, n is ONT squiggle length and o is number of 16Ses to compare. It sort of works with the R9 data I downloaded in an offline manner. I would like to integrate it with ReadUntil and see how much 16S enrichment I can achieve with a mixed uncultured sample. But the Developer Agreement seems to be not friendly to my situation.

I think FPGA in the end can solve some problem. But in most cases, it is better for us to have a good algorithm before the FPGA-ization.