Read Until on a flongle

[looselab]

Read Until on a flongle

[mattloose]

Hi Sam,

Use case on flongle is interesting - you are doing roughly the right thing but I'm going to make some suggestions.

1. The workers flag doesn't help you at present - this is legacy and you can just ignore it and run on a single thread. The heavy lifting here is done by guppy!
2. You are correct - it should be 126 channels and not 128 for a flongle.
3. I presume you are using our bulk file - this means that the total data will come from 512 channels and so you need to parse out the results from channels 1-128 and compare with the rest of the flowcell to see what's actually happening.
4. Can you share the toml file you used - it looks to me as though it is the same as our original - it should be selecting chromosome 21 and 22 only. To look at this on playback we plot the mean read length of reads mapped to each chromosome - you should see a difference on channels 1-128 compared with the rest of the flowcell.

Performance wise, you may find that the chunk size setting of 0.4 is not helping you with the CPU basecaller - it may be more performant to set a chunk size of 1 second and test this.

Hope some of this helps - feel free to post more on this thread with clearer analysis of the output.

Cheers

Matt

[SamMcGreig]

Hi Matt,

Thanks for the quick response. I've made changes from points 1 + 2 and will tests them again this week.

Yes, I am using your bulk file. I will parse the results from channels 1-128 and see what happens to them.

I'll let you know how the testing goes and be in touch with results.

toml file:
[caller_settings]
config_name = "dna_r9.4.1_450bps_fast"
host = "127.0.0.1"
port = 5556

[conditions]
reference = "/ReadUntil/read_until/human_genome/human.mmi"

[conditions.0]
name = "select_chr_21_22"
control = false
min_chunks = 0
max_chunks = 4
targets = ["NC_000021.9", "NC_000022.11"]
single_on = "stop_receiving"
multi_on = "stop_receiving"
single_off = "unblock"
multi_off = "unblock"
no_seq = "proceed"
no_map = "proceed"

[SamMcGreig]

Hi Matt,

Just updating with some results:
Control (channels 127+)

Read Until (channels 1-126)

Looks like it's worked well for one chr22 and not as well for chr21 - but overall I think it's worked quite well. Hopefully we'll be able to begin testing on our own samples in the near future.

Any feedback/suggestions would be welcome
Sam