Rohan Gala

Researcher @alleninstitute, by way of @physicsneu and @iitbombay

[Rohan Gala @rhngla]

@jmschreiber91 This looks to be related to over-smoothing and over-squashing problems explored in graph neural networks, e.g. https://t.co/5FBjSr8nUn

[Rohan Gala @rhngla]

Applications for the Allen Institute summer internship program are now open!

Help us think about and build computational tools towards discovering neuronal communication networks using spatial transcriptomics datasets.

Apple here! https://t.co/2lMrKBP41h

[Rohan Gala @rhngla]

@SuryaGanguli @anshulkundaje @rhngla - I work at Allen Institute, on applied ML for large scale neuroscience and -omics experiments.

[Rohan Gala @rhngla]

@nirsd DM’d you!

[Su-In Lee @suinleelab]

@NatureComms just published the work of my outstanding @uwcse student @ianccovert and our amazing collaborators at @AllenInstitute and @HHMIJanelia. Thank you so much, Uygar Sümbül, @rhngla,Tim Wang, and @svoboda314, for working with us! Congrats all!
https://t.co/iUa6YSZV9F

[Rohan Gala @rhngla]

I prompted DreamStudio with “A modern scientist using machines to study organization of the brain, in the style of Raja Ravi Verma” to generate the initial image.

I then used Dall-E to inpaint a steampunk engine, followed by some cleanup using photoshop.

[Rohan Gala @rhngla]

Of minds and machines, in the style of Raja Ravi Verma

[Rohan Gala @rhngla]

@metricausa @AllenInstitute @CGrienberger @AToliasLab @MarkCembrowski @neuronJoy @NatComputSci We developed coupled autoencoders for this
https://t.co/9AItxBWKNq

Similar approaches have since been developed to analyze other partially-paired single cell resolution multimodal datasets (e.g. BABEL @james_y_zou, scGLUE @gao_laboratory, MultiVI @YosefLab)

[Rohan Gala @rhngla]

Presenting our development of coupled autoencoders to integrate multi-modal single cell data for cell type analysis at #NeurIPS2019 tomorrow:
--> Poster 145, Dec 11 (Wed), 10:45 am
Paper: https://t.co/iYd7RaHHpG

[Rohan Gala @rhngla]

@metricausa @AllenInstitute @CGrienberger @AToliasLab @MarkCembrowski @neuronJoy Uncovering cross-modal associations, defining cell types identifiable across modalities, mapping new data to atlases e.g. ones generated by @AllenInstitute and @AToliasLab remain challenging computational problems.

Our @NatComputSci work on this topic: https://t.co/B7jXCSYVWJ

[Rohan Gala @rhngla]

@jeffgilchrist @ShivenTaneja Cool thread! I had thought about it from the perspective of mask reuse and wrote a summary here: https://t.co/UHk7wIrExT

[Rohan Gala @rhngla]

I caught the Seattle freeze outside my window today #wawx #seattlewx

[Rohan Gala @rhngla]

@michael_nielsen Same! My graduate school lab was a 5 min walk away @mfaboston, and it was fun to work from there every now and then.

[Rohan Gala @rhngla]

PROPOSE also enables gene selection using multi-modal data - the selected genes can recover more than just cell type labels. It’s been a pleasure to collaborate on this work!

[Su-In Lee @suinleelab]

Can we use RNA-seq data to select an effective probe gene set for spatial transcriptomics? We answer this question in our study (Cover et al. @ianccovert in collaboration with Uygar Sümbül
@AllenInstitute)
https://t.co/mICkJybtei

[Rohan Gala @rhngla]

@dyfbrain @patrickmineault / for beginners; Also, this birds’ eye view of python is a handy reference to just glance through once in a while : https://t.co/adRHD3nlcG

[Rohan Gala @rhngla]

@dyfbrain @patrickmineault Once familiar with the concept of environments, they can use this repo to understnad how to reuse code locally: https://t.co/3f6ROQ8kWn. Understanding different paths, and getting to the point of being able to import self-written functions in notebooks/scripts is helpful

[Rohan Gala @rhngla]

@gottapatchemall That looks serene!

[Rohan Gala @rhngla]

@neuroecology This recent paper proposes credit assignment mediated by neuromodulators as the computational role for cell types : https://t.co/i41nIyo2G1

[David Feng @dyfbrain]

(1/3) Scientific Computing at the Allen Institute for Neural Dynamics is hiring!

1: Scientific Data Engineer: design and build systems to manage scientific metadata and automate processing pipelines. https://t.co/bfrJXKQ1FO

[Rohan Gala @rhngla]

@davisidarta https://t.co/bf4jHW0uHM - lots of cells, multiple technologies, rich metadata.