BOL: Related items

NVIDIA and Arc Institute Unveil Evo 2: A Breakthrough AI for DNA Design

BioStar — Fri, 21 Feb 2025 10:39:47 -0600

NVIDIA and the Arc Institute have introduced Evo 2, a groundbreaking AI model designed to understand, predict, and generate DNA sequences. This marks a major advancement in computational biology, offering scientists an unprecedented tool to decode the genetic blueprint of life and even design entirely new biological systems.

The Power of Evo 2: AI Meets DNA

Evo 2 is the largest AI model for biology ever created, trained on an astonishing 9.3 trillion DNA "letters" (nucleotides) carefully selected from genomes spanning the entire tree of life. This massive dataset ensures that Evo 2 can recognize patterns and relationships in genetic sequences at an unparalleled scale.

For the first time, scientists can design DNA with AI, moving beyond simple sequence analysis to active DNA generation. Evo 2 enables researchers to predict, modify, and even create entire genetic sequences, opening new possibilities in medicine, agriculture, and synthetic biology.

Decoding the Dark Genome

One of the biggest challenges in genetics is understanding the non-coding regions of DNA—vast stretches of the genome that do not code for proteins but play crucial roles in regulating gene expression. These regions control when and how genes are activated, influencing everything from development to disease.

Evo 2 is designed to decode these non-coding elements, helping researchers uncover their functions and use this knowledge to develop gene-based therapies, synthetic life forms, and precision agriculture solutions.

From Reading DNA to Writing It

To put Evo 2’s impact into perspective:

Previous AI models could "read" DNA like a book, analyzing genetic sequences and identifying patterns.
Evo 2 can "write" entirely new DNA, designing functional genes, chromosomes, and even full genomes from scratch.

This means scientists can now engineer biological systems with AI, designing new proteins, metabolic pathways, and genetic circuits to address real-world challenges.

A Step Toward Generative Biology

The Arc Institute describes Evo 2 as a major step toward "generative biology"—a revolutionary approach where AI is used to create novel biological structures rather than just analyzing existing ones. This could lead to breakthroughs such as:

New medicines: AI-generated enzymes and proteins tailored for targeted therapies.
Disease-resistant crops: Genetically optimized plants for higher yield and climate resilience.
Synthetic organisms: Custom-designed microbes for bioremediation, biofuel production, and industrial applications.

An Open-Source Revolution

Unlike many proprietary AI models, Evo 2 is open source, making its capabilities accessible to researchers worldwide. This democratization of AI-driven biology means that scientists from different disciplines can collaborate, experiment, and innovate, accelerating discoveries in genetic engineering and synthetic biology.

With Evo 2, the boundaries of what’s possible in DNA design, genetic engineering, and biological innovation are being redrawn. The future of life sciences is no longer just about understanding life’s code—it’s about writing it.

Aurora: A New Detective for Bacterial Genomes

BioStar — Fri, 21 Aug 2026 11:31:00 -0500

Imagine trying to solve a mystery with thousands of clues—but some of the clues are labelled incorrectly.

That is the challenge researchers face when studying how bacteria adapt to different environments. A bacterial gene may appear to be linked to a particular habitat, when the real reason is simply that closely related bacteria happen to live there.

In their 2025 Genome Biology paper, Bujdoš, Walter, and O’Toole introduce aurora, a machine-learning tool designed to tackle this problem.

Aurora identifies potentially mislabeled or unusual bacterial strains before performing genome-wide association studies (GWAS). By cleaning up the dataset and accounting for bacterial evolutionary relationships, it can help researchers find genetic features that are more genuinely connected to habitat adaptation.

The researchers tested aurora using simulated and real bacterial datasets and found that it could recover important genetic associations even when datasets contained misleading labels.

The bigger lesson is simple: better biological discoveries often begin with better data.

Aurora gives researchers a new way to separate real genetic clues from misleading ones—and could help us better understand how bacteria adapt, survive, and evolve in the environments they call home.

Based on Bujdoš et al., “aurora: a machine learning GWAS tool for analyzing microbial habitat adaptation,” Genome Biology (2025).[Read the original paper](https://link.springer.com/article/10.1186/s13059-025-03524-7?)

Luo Lab | Symbiosis Genomics & Evolution

Wed, 09 Sep 2026 03:30:30 -0500

We study the evolutionary genomics of marine invertebrates to understand their origins and diversity. Our lab combines high-throughput sequencing and single-cell transcriptomics to explore a wide range of non-model systems. We are particularly interested in how evolutionary novelty arises, with a focus on animal development and photosymbiosis.

Research Directions

Stony corals: evolution of novelty and photosymbiosis

Symbiotic acoels: cell type evolution and photosymbiosis

Animal genomes: structural evolution and gene regulation

https://sgel.biodiv.tw/home

k-mers tutorial - classification and taxonomy

Neel — Thu, 26 Aug 2021 10:28:43 -0500

DNA k-mers underlie much of our assembly work, and we (along with many others!) have spent a lot of time thinking about how to store k-mer graphs efficiently, discard redundant data, and count them efficiently.

More recently, we've been enthused about using k-mer based similarity measures and computing and searching k-mer-based sketch search databases for all the things.

But I haven't spent too much talking about using k-mers for taxonomy, although that has become an ahem area of interest recently, if you read into our papers a bit.

In this blog post I'm going to fix this by doing a little bit of a literature review and waxing enthusiastic about other people's work. Then in a future blog post I'll talk about how we're building off of this work in fun! and interesting? ways!

Address of the bookmark: http://ivory.idyll.org/blog/2017-something-about-kmers.html

npScarf: real-time scaffolder using SPAdes contigs and Nanopore sequencing reads

Shruti Paniwala — Mon, 11 Jun 2018 05:14:57 -0500

npScarf (jsa.np.npscarf) is a program that connect contigs from a draft genomes to generate sequences that are closer to finish. These pipelines can run on a single laptop for microbial datasets. In real-time mode, it can be integrated with simple structural analyses such as gene ordering, plasmid forming.

Address of the bookmark: http://japsa.readthedocs.io/en/latest/tools/jsa.np.npscarf.html

BIGMAC : breaking inaccurate genomes and merging assembled contigs for long read metagenomic assembly

Jit — Mon, 22 May 2017 05:43:51 -0500

This tool is for users to upgrade their metagenomics assemblies using long reads. This includes fixing mis-assemblies and scaffolding/gap-filling. If you encounter any issues, please contact me at kklam@eecs.berkeley.edu. My name is Ka-Kit Lam.

https://github.com/kakitone/MetaFinisherSC

https://github.com/kakitone/BIGMAC

Address of the bookmark: https://github.com/kakitone/BIGMAC

PERGA: A Paired-End Read Guided De Novo Assembler for Extending Contigs Using SVM and Look Ahead Approach

Rahul Nayak — Tue, 05 Jun 2018 09:57:11 -0500

PERGA - Paired End Reads Guided Assembler PERGA is a novel sequence reads guided de novo assembly approach which adopts greedy-like prediction strategy for assembling reads to contigs and scaffolds. Instead of using single-end reads to construct contig, PERGA uses paired-end reads and different read overlap sizes from O ≥ Omax to Omin to resolve the gaps and branches. Moreover, by constructing a decision model using machine learning approach based on branch features, PERGA can determine the correct extension in 99.7% of cases. PERGA will try to extend the contigs by all feasible nucleotides and determine if these multiple extensions due to sequencing errors or repeats by using looking ahead technology, and it also try to separate the different repeats of nearby genomic regions to make the assembly result more longer and accurate. The simulated E.coli paired-end reads data are generated using GemSim (KE McElroy, F Luciani, T Thomas. Gemsim: General, Error-Model Based Simulator of Next-Generation Sequencing Data. BMC Genomics 2012, 13:74), with coverage 50x, 60x, 100x, read lengths 100-bp, and can be downloaded from https://github.com/zhuxiao/data_PERGA.

Address of the bookmark: https://github.com/hitbio/PERGA

gfastats: The swiss army knife for genome assembly.

Abhi — Thu, 08 Sep 2022 06:03:05 -0500

gfastats is a single fast and exhaustive tool for summary statistics and simultaneous *fa* (fasta, fastq, gfa [.gz]) genome assembly file manipulation. gfastats also allows seamless fasta<>fastq<>gfa[.gz] conversion. It has been tested in genomes even >100Gbp.

Address of the bookmark: https://github.com/vgl-hub/gfastats

Public Databases for Bioinformatics !

Jit — Tue, 23 Mar 2021 05:32:15 -0500

https://www.nature.com/articles/s41467-020-17155-y

Server Infrastructure:

File Server:

dhara: Synology 3614 Storage Appliance
4 Core Xeon
108TB disk storage
10Gb ethernet to SCG3
Access atx: dhara:5000
Has btsync server (try it - its much better than dropbox)

Compute Servers:

nandi: Kundaje and Phi Server
24 intel cores
256GB RAM
500GB of SSD storage 
36TB RAID6 local storage
4 Intel Phi's (space for 4 more GPU's)


durga: Montgomery and sensitive data
24 intel cores
256GB RAM
500GB of SSD RAID0 storage 
60TB RAID6 local storage

mitra: Bassik and Web/DB Server
24 core
256GB RAM 
500GB of SSD RAID0 storage 
36TB RAID6 local storage

vayu: Kundaje GPU server
4 core
64GB RAM 
200GB of SSD storage 
8TB RAID10 local storage
4 Nvidia GTX 970 4GB GPUs

amold: Bickel and SGE server
32 AMD core
128GB RAM 
200GB of SSD storage 
12TB RAID5 local storage

wotan: Bickel and SGE server
64 AMD core
256GB RAM 
200GB of SSD storage 
12TB RAID5 local storage

Filesystem:

/users/$USER
default home directory
full backups nightly 
nfs mount to dhara
should store code, papers, and other highly processed data here

/mnt/data/
globally accessible data
should store common data here
e.g. genomes and indexes, annotations, ENCODE data  
if you dont want this to count towards your quote you must chown

/mnt/lab_data/$LAB/
lab accessible data
should store lab project data here 
e.g. ATAC-seq prediction data, enhancer prediction, motif calls

/srv/scratch/$USER
fast local storage
not backed up, but on raid and data will never be deleted
most analysis should be performed here

/srv/persistent/$USER
fast local storage
synced nightly, but not backed up
       ie if the hard drives fail or you delete something and notice 
       within 24 hours we can recover. Otherwise not. (vs home which is 
       properly backed up )  
intermediate analysis products that would be hard to recover should be stored here 
       e.g. stochastic analysis results that need to be kept so that paper 
       results can be reproduced

/srv/www/$LABNAME/
web accessible from mitra.stanford.edu
*NOT BACKED UP*

Some parallel programming patterns:

# gzip a bunch of files
parallel gzip -- *.FILESTOGZIP

# fork example in python:
(for more detailed examples look at 
 https://github.com/nboley/grit/ grit/lib/multiprocessing_utils.py)

import os
import time
import random

import multiprocessing

class ProcessSafeOPStream( object ):
    def __init__( self, writeable_obj ):
        self.writeable_obj = writeable_obj
        self.lock = multiprocessing.Lock()
        self.name = self.writeable_obj.name
        return
    
    def write( self, data ):
        self.lock.acquire()
        self.writeable_obj.write( data )
        self.writeable_obj.flush()
        self.lock.release()
        return
    
    def close( self ):
        self.writeable_obj.close()

def worker(queue, ofp):
    # Try without this
    random.seed()
    while True:
        i = queue.get()
        if i == 'FINISHED': return
        # simulate an expensive function
        x = random.random()
        time.sleep(x/10)
        print i, x
        ofp.write("%i\t%s\n" % (i, x))

NSIMS = 10000
NPROC = 25

# populate queue
todo = multiprocessing.Queue()
for i in xrange(NSIMS): todo.put(i)
for i in xrange(NPROC): todo.put('FINISHED')

ofp = ProcessSafeOPStream( open("output.txt", "w") )

pids = []
for i in xrange(NPROC):
    pid = os.fork()
    if pid == 0:
       worker(todo, ofp)
       os._exit(0)
    else:
       pids.append(pid)  

for pid in pids:
    os.waitpid(pid, 0)

ofp.close()

print "FINISHED"

For use case 1 we obtained the following ENCODE and ROADMAP datasets https://www.encodeproject.org/files/ENCFF446WOD/@@download/ENCFF446WOD.bed.gz, https://www.encodeproject.org/files/ENCFF546PJU/@@download/ENCFF546PJU.bam, https://www.encodeproject.org/files/ENCFF059BEU/@@download/ENCFF059BEU.bam. Blacklisted regions were obtained from http://mitra.stanford.edu/kundaje/akundaje/release/blacklists/hg38-human/hg38.blacklist.bed.gz. The human genome version hg38 was obtained from http://hgdownload.cse.ucsc.edu/goldenPath/hg38/bigZips/hg38.fa.gz.

For use case 2 we used the set of narrowPeak files summarized in https://github.com/wkopp/janggu_usecases/tree/master/extra/urls.txt (archived version v1.0.1). The human genome version hg19 was obtained from http://hgdownload.cse.ucsc.edu/goldenPath/hg19/bigZips/hg19.fa.gz

For use case 3 we used the ENCODE datasets https://www.encodeproject.org/files/ENCFF591XCX/@@download/ENCFF591XCX.bam, https://www.encodeproject.org/files/ENCFF736LHE/@@download/ENCFF736LHE.bigWig, https://www.encodeproject.org/files/ENCFF177HHM/@@download/ENCFF177HHM.bam as we as the GENCODE annotation v29 from ftp://ftp.ebi.ac.uk/pub/databases/gencode/Gencode_human/release_29/gencode.v29.annotation.gtf.gz.

Address of the bookmark: http://mitra.stanford.edu/

Ribbon: Visualizing complex genome alignments and structural variation:

Jit — Wed, 29 Nov 2017 07:40:22 -0600

Ribbon can be used for long reads, short reads, paired-end reads, and assembly/genome alignments. Instructions for each data format are available by clicking on "instructions" in each tab on the right.

Local installation:

You can install Ribbon locally from Github by following the instructions here: https://github.com/MariaNattestad/Ribbon

Address of the bookmark: http://genomeribbon.com/