jonathangoeke.bsky.social
Group Leader, Genome Institute of Singapore, A*STAR, Working on computational methods for long read RNA-Seq
https://github.com/goekelab
17 posts
120 followers
132 following
Regular Contributor
Conversation Starter
comment in response to
post
Learn more about the team led by @jonathangoeke.bsky.social and their work at:
www.a-star.edu.sg/gis/news-eve...
comment in response to
post
(reposted with correct URL)
comment in response to
post
This work was down by Andre Sim, Min Hao Ling, Chen Ying, Sui Yue, together with Jay Shin's team at GIS. Let us know your comments and feedback, we aim to improve Bambu with every release! github.com/GoekeLab/bambu
comment in response to
post
The recommended way to run bambu-clump is through the nextflow pipeline (bambu-pipe), which optimises barcode detection, alignment, transcript discocery and quantification, and it also includes an option to identify fusion transcripts in single cells github.com/GoekeLab/bam...
comment in response to
post
...reads are only assigned to transcripts after similar cells are clustered into groups, reducing the noise in transcript expression estimates due to missing data. We also provide unique counts for single cells (and EM counts for genes/experiments with high sequencing depth)
comment in response to
post
The challenge with transcript expression estimation is the limited number of reads per gene, which makes the assignment of (non unique) reads to transcripts difficult. To address this, Bambu performs a clustered EM, in which ....
comment in response to
post
The key challenges were dimensionality (>200k transcripts x thousands cells) and estimation of transcript expression. The dimensionality we handle by optimising the representation of reads in memory using barcoded read classes, making transcript discovery and read-to-transcript assignment very fast
comment in response to
post
This work was down by Andre Sim, Min Hao Ling, Chen Ying, Sui Yue, together with Jay Shin's team at GIS. Let us know your comments and feedback, we aim to improve Bambu with every release! github.com/GoekeLab/bambu
comment in response to
post
The recommended way to run bambu-clump is through the nextflow pipeline (bambu-pipe), which optimises barcode detection, alignment, transcript discocery and quantification, and it also includes an option to identify fusion transcripts in single cells github.com/GoekeLab/bam...
comment in response to
post
...reads are only assigned to transcripts after similar cells are clustered into groups, reducing the noise in transcript expression estimates due to missing data. We also provide unique counts for single cells (and EM counts for genes/experiments with high sequencing depth)
comment in response to
post
The challenge with transcript expression estimation is the limited number of reads per gene, which makes the assignment of (non unique) reads to transcripts difficult. To address this, Bambu performs a clustered EM, in which ....
comment in response to
post
The dimensionality we handle by optimising the representation of reads in memory using barcoded read classes, which allows fast transcript discovery and read-to-transcript assignment