#p2solo — Public Fediverse posts

@nanopore Updated: GPU/MinKNOW basecalling server walkthrough:

https://gringer.gitlab.io/presentation-notes/2021/10/08/gpu-calling-in-minknow/

#P2Solo

@nanopore Updated: GPU/MinKNOW basecalling server walkthrough:

https://gringer.gitlab.io/presentation-notes/2021/10/08/gpu-calling-in-minknow/

#P2Solo

@nanopore We're running another lot of cDNA samples on our P2 Solo. Flow cell loading and available pores looks great. Not quite a full sea of green, but that's to be expected from an older PromethION flow cell. They seem to be more time-sensitive than the MinION and Flongle flow cells.

#P2Solo

@nanopore We're running another lot of cDNA samples on our P2 Solo. Flow cell loading and available pores looks great. Not quite a full sea of green, but that's to be expected from an older PromethION flow cell. They seem to be more time-sensitive than the MinION and Flongle flow cells.

#P2Solo

@nanopore Super-accuracy basecalling is done for our most recent P2 Solo runs. I'm seeing about q20 modal accuracy from mapped reads.

#p2Solo #cDNA #nanopore

@nanopore Super-accuracy basecalling is done for our most recent P2 Solo runs. I'm seeing about q20 modal accuracy from mapped reads.

#p2Solo #cDNA #nanopore

@nanopore we actually crossed over 70M reads with one of the flow cells!

#P2Solo #cDNASequencing #nanopore

@nanopore we actually crossed over 70M reads with one of the flow cells!

#P2Solo #cDNASequencing #nanopore

@nanopore I now have a comparison of the same sample on Flongle (cleanup using SFB) and PromethION (cleanup using LFB) flow cells.

These were both called with the new bacterial caller in simplex mode. According to LAST-train, the qScore-adjusted substitution accuracy for the Flongle was q41, and the qScore-adjusted substitution accuracy for the PromethION was q45.

#P2Solo #DNAladder

@nanopore I now have a comparison of the same sample on Flongle (cleanup using SFB) and PromethION (cleanup using LFB) flow cells.

These were both called with the new bacterial caller in simplex mode. According to LAST-train, the qScore-adjusted substitution accuracy for the Flongle was q41, and the qScore-adjusted substitution accuracy for the PromethION was q45.

#P2Solo #DNAladder

@nanopore hmm... looks like I messed up on the 3kb and 1.5kb ladder bands.

This new kmer-based mapper seems to be doing the right things, though. I'm very pleased.

#DNAladder #P2Solo

@nanopore hmm... looks like I messed up on the 3kb and 1.5kb ladder bands.

This new kmer-based mapper seems to be doing the right things, though. I'm very pleased.

#DNAladder #P2Solo

@[email protected] That new bacterial @nanopore bascalling model is doing pretty well.

#DNAladder #P2Solo

@[email protected] That new bacterial @nanopore bascalling model is doing pretty well.

#DNAladder #P2Solo

#P2Solo Data upload to our experimental archive. 641.15 GB of raw pod5 files. This was the lower end of acceptable data output for a P2 Solo run (about 50 Gb of sequence data).

#P2Solo Data upload to our experimental archive. 641.15 GB of raw pod5 files. This was the lower end of acceptable data output for a P2 Solo run (about 50 Gb of sequence data).

@nanopore #P2Solo another cDNA run that has exceeded 25M reads. The computer froze [again] in the middle of a sequencing run, after getting 6.24M reads, but I now have >25M reads from after the restart.

[FWIW, I care about 25M because that's the maximum that could ever be generated from a single Revio run, assuming perfect performance from all ZMWs]

@nanopore #P2Solo another cDNA run that has exceeded 25M reads. The computer froze [again] in the middle of a sequencing run, after getting 6.24M reads, but I now have >25M reads from after the restart.

[FWIW, I care about 25M because that's the maximum that could ever be generated from a single Revio run, assuming perfect performance from all ZMWs]

@nanopore Canu-corrected #P2Solo read, demonstrating qInf (i.e. 100% accuracy)

@nanopore Canu-corrected #P2Solo read, demonstrating qInf (i.e. 100% accuracy)

#P2Solo @nanopore Currently sequencing *Nippostrongylus brasiliensis* on a PromethION flow cell, prepared using LSK114 ligation sequencing kit (using ligase enzyme that is about 2 years past its claimed expiry date).

We're about 50 minutes into the run, and it has covered the 300~400Mb genome about 5 times.

I'm hoping to get coverage above 30X before flow cell degredation sets in. If, against the odds, the flow cell survives for a long time, I might get enough reads for a population analysis.

#P2Solo @nanopore Currently sequencing *Nippostrongylus brasiliensis* on a PromethION flow cell, prepared using LSK114 ligation sequencing kit (using ligase enzyme that is about 2 years past its claimed expiry date).

We're about 50 minutes into the run, and it has covered the 300~400Mb genome about 5 times.

I'm hoping to get coverage above 30X before flow cell degredation sets in. If, against the odds, the flow cell survives for a long time, I might get enough reads for a population analysis.

@nanopore #P2Solo digital electrophoresis plot showing estimated tail lengths from various cDNA samples (bottom two lines are after two PCR cycles; all the others are after 14 PCR cycles).

@nanopore #P2Solo digital electrophoresis plot showing estimated tail lengths from various cDNA samples (bottom two lines are after two PCR cycles; all the others are after 14 PCR cycles).

#P2Solo @nanopore second trial of rapid PCR cDNA sequencing, running 18 samples on a PromethION P2 Solo flow cell.

[Our first trial was on a MinION flow cell]

I'm hoping to get over 25M reads from this run, which should be easily achievable given that it should get to 2M reads within the first hour, and it'll be running for 72 hours.

My current estimate is for about 70M reads by the end of the run.

#P2Solo @nanopore second trial of rapid PCR cDNA sequencing, running 18 samples on a PromethION P2 Solo flow cell.

[Our first trial was on a MinION flow cell]

I'm hoping to get over 25M reads from this run, which should be easily achievable given that it should get to 2M reads within the first hour, and it'll be running for 72 hours.

My current estimate is for about 70M reads by the end of the run.

#P2Solo #LadderSequencing @nanopore I have my first band (100% identity match to lots of different vectors):

>band_100bp
TGCACGAACCCCCCGTTCAGCCCGACCGCTGCGCCTTATCCGGTAACTATCGTCTTGAGTCCAACCCGGTAAGACACGACTTATCGCCACTGGCAGCAGCCACTGGTAACAGGATTAGCAGAGCGAGGTATGTAGGC

It's 137bp, rather than 100bp

#P2Solo #LadderSequencing @nanopore I have my first band (100% identity match to lots of different vectors):

>band_100bp
TGCACGAACCCCCCGTTCAGCCCGACCGCTGCGCCTTATCCGGTAACTATCGTCTTGAGTCCAACCCGGTAAGACACGACTTATCGCCACTGGCAGCAGCCACTGGTAACAGGATTAGCAGAGCGAGGTATGTAGGC

It's 137bp, rather than 100bp

#P2Solo @nanopore Gel vs sequencer

Peak heights are a little different; I'm hoping that some of that will be fixed up after doing a direct mapping to the sequences, rather than looking at all reads.

#P2Solo @nanopore Gel vs sequencer

Peak heights are a little different; I'm hoping that some of that will be fixed up after doing a direct mapping to the sequences, rather than looking at all reads.

@nanopore #P2Solo this is what needle shearing does to human DNA

@nanopore #P2Solo this is what needle shearing does to human DNA

#P2Solo just looking at a few @nanopore lambda reads, called using dorado stereo duplex with model '[email protected]'. Local Mapping with LAST/RY4, looking at gap-compressed identity.

Median accuracy: 99.89% (q30)
Maximum accuracy: 99.98% (q37)

That maximum accuracy read spans 99.70% of the lambda genome (i.e. 48355 bp), and has 11 errors within that span.

#P2Solo just looking at a few @nanopore lambda reads, called using dorado stereo duplex with model '[email protected]'. Local Mapping with LAST/RY4, looking at gap-compressed identity.

Median accuracy: 99.89% (q30)
Maximum accuracy: 99.98% (q37)

That maximum accuracy read spans 99.70% of the lambda genome (i.e. 48355 bp), and has 11 errors within that span.

#P2Solo Gosh, that @nanopore duplex basecalling is accurate! Here's a *single base read* which has an accuracy of around q33 (99.95% accuracy):

@4b2e1d20-f050-405a-bc7f-3f6d7cf2a41f;134866c0-cd76-4da1-b717-ab2caf260b9c
C
+
B

#P2Solo Gosh, that @nanopore duplex basecalling is accurate! Here's a *single base read* which has an accuracy of around q33 (99.95% accuracy):

@4b2e1d20-f050-405a-bc7f-3f6d7cf2a41f;134866c0-cd76-4da1-b717-ab2caf260b9c
C
+
B

#P2Solo @nanopore run finished. Using PacBio's website figures for Revio, and assuming *all* ZMWs are working, I get the following costs:

Per million reads:

Revio: $40
Novaseq S4: $2.75
P2 Solo: $22

Per gigabase:

Revio: $11
Novaseq S4: $10
P2 Solo: $12.50

[this is assuming the most expensive P2 Solo flow cell purchase]

#P2Solo @nanopore run finished. Using PacBio's website figures for Revio, and assuming *all* ZMWs are working, I get the following costs:

Per million reads:

Revio: $40
Novaseq S4: $2.75
P2 Solo: $22

Per gigabase:

Revio: $11
Novaseq S4: $10
P2 Solo: $12.50

[this is assuming the most expensive P2 Solo flow cell purchase]

Note for future reference: @nanopore #P2Solo is mostly dead once its sequencing pore count drops down to that of a MinION flow cell.

The flow cell voltage is at -230 mV, so I'm expecting that the death will be a bit quicker.

Still, it's impressive to see it still fighting after 56 hours.

Note for future reference: @nanopore #P2Solo is mostly dead once its sequencing pore count drops down to that of a MinION flow cell.

The flow cell voltage is at -230 mV, so I'm expecting that the death will be a bit quicker.

Still, it's impressive to see it still fighting after 56 hours.

#P2Solo @nanopore Sequenced base density curve for DNA ladder (1µl) + DCS (1µl) + lambda (19µl) on a P2 Solo. This is only 6 M of the ~43M reads [so far] from the run. I'm hesitant to collect more length data because the reason it's only 6 M is that's how many reads I was able to get statistics on before the computer decided it was being overworked and spontaneously rebooted.

https://assets.fishersci.com/TFS-Assets/LSG/manuals/MAN0000843_trackit-1kb-plusDNAladder_PI.pdf

#P2Solo @nanopore Sequenced base density curve for DNA ladder (1µl) + DCS (1µl) + lambda (19µl) on a P2 Solo. This is only 6 M of the ~43M reads [so far] from the run. I'm hesitant to collect more length data because the reason it's only 6 M is that's how many reads I was able to get statistics on before the computer decided it was being overworked and spontaneously rebooted.

https://assets.fishersci.com/TFS-Assets/LSG/manuals/MAN0000843_trackit-1kb-plusDNAladder_PI.pdf

#P2Solo @nanopore records for our institute are being broken in less than a day by this first run on the replacement sequencer (third P2 Solo run overall).

8 hours of sequencing, and we've now got our best yield in terms of bases sequenced (previous record was 18 Gb).

As long as the computer doesn't crash on me, I expect we'll easily get past Revio's 25M reads (max) per run by the time I wake up tomorrow.

I look forward to updating PacBio's table.

#P2Solo @nanopore records for our institute are being broken in less than a day by this first run on the replacement sequencer (third P2 Solo run overall).

8 hours of sequencing, and we've now got our best yield in terms of bases sequenced (previous record was 18 Gb).

As long as the computer doesn't crash on me, I expect we'll easily get past Revio's 25M reads (max) per run by the time I wake up tomorrow.

I look forward to updating PacBio's table.