ALL Metrics
-
Views
-
Downloads
Get PDF
Get XML
Cite
Export
Track
Opinion Article
Revised

hackseq: Catalyzing collaboration between biological and computational scientists via hackathon

[version 2; peer review: 2 approved]
PUBLISHED 10 Apr 2017
OPEN PEER REVIEW
REVIEWER STATUS

This article is included in the Bioinformatics Education and Training Collection collection.

This article is included in the Hackathons collection.

Abstract

hackseq (http://www.hackseq.com) was a genomics hackathon with the aim of bringing together a diverse set of biological and computational scientists to work on collaborative bioinformatics projects. In October 2016, 66 participants from nine nations came together for three days for hackseq and collaborated on nine projects ranging from data visualization to algorithm development. The response from participants was overwhelmingly positive with 100% (n = 54) of survey respondents saying they would like to participate in future hackathons. We detail key steps for others interested in organizing a successful hackathon and report excerpts from each project.

Keywords

Hackathon, Genomics, Bioinformatics, Open Science, Diversity in Science

Revised Amendments from Version 1

We added a brief description of the project leaders and participants role during the hackathon to the research project summaries. We also expanded the discussion  of female participation, as suggested by the referees. We also corrected some typos.

To read any peer review reports and author responses for this article, follow the "read" links in the Open Peer Review table.

Introduction

Technological advances in the biological sciences have led to an increasing availability of so-called ‘-omic’ datasets, allowing fundamental questions in biology to be answered at an unprecedented rate1. However, these datasets are complex, requiring novel and specialized informatics tools for proper analysis and to overcome the computational bottleneck in research. Open-source bioinformatics tools and pipeline development accelerates the rate of research by allowing the community to both reuse and thoroughly assess such methods. Thus, by solving biological problems in an open and collaborative manner, the field can progress at a faster rate than if code remains unavailable to the larger community2.

Hackathons offer a solution to catalyze tool and pipeline development for biological data science, as well as foster interdisciplinary collaborations3. These events aim to solve defined computational problems over a period of days by bringing together small teams of individuals with different and diverse skillsets. Although frequently valuable for the outputs they generate, hackathons have faced criticism due to low levels of diversity amongst participants4. We therefore established hackseq, a genomics hackathon collective (http://www.hackseq.com) that aims to promote open science, collaboration and diversity. We placed special emphasis in promoting leadership amongst women, minorities and early-career scientists. The inaugural hackseq event took place over three days in October 2016 in Vancouver (British Columbia, Canada) and was a satellite event to the annual American Society of Human Genetics (ASHG) meeting. Here we report a summary of this hackathon in the hopes of promoting similar events in the future.

hackseq format

hackseq was the first genomics hackathon in Vancouver and was based on the NCBI hackathon format3. Some hackathons can be perceived as high-pressure events exclusive to technically inclined and experienced individuals. We thus took measures to ensure that people of all skill levels and backgrounds were encouraged to apply. We structured hackseq as a three-day event that runs primarily from 8AM – 5PM on the Saturday/Sunday/Monday prior to the 2016 ASHG meeting. The hackseq itinerary is accessible on the hackseq github repository (https://github.com/hackseq/October_2016/blob/master/hackseq_2016_schedule.md).

First, we opened a call for ‘team leaders’ to propose a project and lead a small team at hackseq through social media, such as Twitter, making announcements at the Vancouver Bioinformatics User Group (VanBUG) and bioinformatics.ca, as well as direct email contact with potential leaders. We screened the projects to confirm that their aims and scope would be appropriate for a 72 hour hackathon. For the ten accepted projects we used GitHub as a discussion board, creating issue threads (https://github.com/hackseq/hackseq_projects_2016/issues) for each project, allowing prospective participants to view and discuss the details about each project before applying to join a particular team.

Once we established the ten hackseq projects, we opened the call for participants. Our main goal in recruiting participants was to reach out to a diverse group of individuals and to promote participation of women, minorities and early-career scientists. To this end, we partnered with organizations, such as Society for Canadian Women in Science and Technology (SCWIST) and VanBUG, to attract local participants. To encourage early-career scientist involvement, we contacted undergraduate and graduate-level computational sciences and bioinformatics programs at regional universities. To reach the global scientific audience, we contacted several human genetics societies around the world asking them to email participant application information to their respective mailing lists.

To promote economic diversity and lower the barrier to entry for international participants, we partnered with ASHG to create travel awards based on financial need and/or minority status. hackseq had no registration fee. Lastly, we made announcements on Twitter, Galaxy Project’s events calendar, and various international conferences, such as Bioinformatics Open Source Conference 2016 and the 13th International Congress of Human Genetics, leading up to the hackathon.

In the participant application form, prospective participants ranked the top four projects on which they would like to work. Participants could apply for travel awards by ASHG and request child care, covered by our budget, to promote participation amongst parents. The organizing committee and the team leaders evaluated the applications based on not only their skill levels, but also their interests and passion for genomics. To ensure well-rounded teams, we considered both the project preferences and skill levels during the team assignment phase, ensuring a balance between novices and expert coders, and biological and computational expertise. All forms developed for hackseq are available online (https://github.com/hackseq/October_2016/blob/master/Forms.md).

By defining the projects and teams beforehand, participants got to know their team members and prepare technical infrastructure. Teams hit the ground running on the first day, beginning work unprompted by the organizers at 8AM of the first day.

hackseq had 66 participants in attendance from nine nations, divided into nine teams ranging from 3 to 10 individuals. Of the accepted ten projects, two team leaders withdrew prior to the hackathon for personal reasons, and one popular project split into two teams, resulting in nine teams. The mode age-category was 30–34 years old (62.5%) for team leaders, and 25–29 years old (58%) for participants (Figure 1A). Graduate students made up the largest fraction of participants with 48.2%, followed by academic staff (15.5%), industrial scientists (13.8%), undergraduates (10.3%), postdoctoral fellows (6.9%) and academic faculty (5.2%). Notably, the team leaders were more likely to be industry scientists (44.4%) or young faculty (22.2%) (Figure 1B). In total, 22 of 62 (35.5%) participants identified as female and 40 as male. A total of 41% self-identified as Caucasian, 40% as Asian or Pacific Islander, and 19% as Arab, Latin American or unspecified (Figures 1C and D).

b5411d73-9bbb-41da-93a3-8a6ebff9b3c4_figure1.gif

Figure 1. Participant diversity at hackseq 16.

To measure diversity of hackseq participants, we asked team leaders and participants to self-report their (A) age, (B) current occupation, (C) ethnicity and (D) gender. Data is shown for team leaders (yellow) and participants (blue).

CategoryAll ResponsesTeam Leaders
GenderMale407
Female222
Age20-2450
25-29301
30-34135
35-3982
40-4910
50+10
OSOSX38na
Linux11na
Windows10na
OccupationUndergraduate Student60
Graduate Student282
Post-Doc40
Academic Staff91
Academic Faculty32
Industrial Scientist84
EthnicityCaucasian26na
Arab2na
Asian / Pacific Islander25na
Latin American1na
Dataset 1.hackseq demographics.
De-identified demographic data from hackseq participants in the pre-meeting survey/confirmation of attendance.
(A-Z sorted)-Rate your subjective experience at hackseq (not column sorted)---------Please rate the usefulness of the various computational infrastructure (1 = no use, 5 = critical)---
Please write three single word adjectives to describe your experience at hackseq ( comma-delimited plz )How did you find the 3-day (8h x 3) format of the hackathon?Hostile(1)-Collaborative(5)Boring(1)-Fun(5)Uninformative(1)-Enlightening(5)Chaotic(1)-Organized(5)What can we do at hackseq to improve the experience of future participants?What about hackseq did you enjoy the most?Would you participate in an event like hackseq again in the future?Would you be interested in helping organize a future hackseq event?Which programming languages and tools did you and your team use during the course of hackseq? (Comma delimited please)Briefly describe what were the computational requirements (e.g., AWS, ORCA, etc) for you during hackseq.ORCAAmazon AWSOn-site volunteersWhich resource(s) did you wish you had access to that was/were not provided?
Amazing, fun,knowledgeableToo Long4422Beer on siteBeing introduced to new software for analysis to read up upon.YesNo thanksBash, python ORCA311
Amazing, intense, funJust right4454Better compute options (ORCA was problematic at first)collaboratingYesNo thankspython; bash; bwa; GATK; platypus; ORCA - would have been helpful to have more local HPC - AWS is difficult for working with large datasets. 513More high memory nodes.
Awesome! Just right5555Better heating, trash cansCollaborating with a diverse groupYesNo thanksBash and pythonAWS and ORCA454More time from Shaun
collaborative, fun, excitingJust right3542Better integration for presentations - I felt bad missing some of them since i was in the middle of working, maybe have them first thing or last thing in the day to get more participationCollaborationYesNo thanksR, python, bash scripting324
collaborative,friendly,stimulatingJust right5555Better Server availability for computeCollaboration among team members.YesNo thankspython,platypus,linux,bwa,samtoolsWe used ORCA, but it was horribly slow and unresponsive, so we instead utilized our own HPCs for analysis. Didn't have time to learn and use the AWS.115A better version of ORCA, WestGrid?
collaborative,fun,enrichingJust right4453Better support for team leadersCollaboration, good spritYesNo thanksR, python, various software/packagesFairly minimal for our project; we did it all on our laptops.113
Collaborative,humbling,funJust right4544can not think of right nowCollaborative atmosphereYesNo thankspython, bash, c++, BLASTAWS, ORCA355
Collaborative,synergy,nerdyJust right5555Confusing to have posts at multiple places on GitHub, and on Slack. It would be helpful to have team leaders be able to have input on selecting team members. connecting with people and have funYesNo thankspython, R, bedtools, bash213
Diverse, enthusiastic, lighthearted Just right5445don't know yetEveryone in my team was very collaborative and we worked great together. YesNo thanksR, python, command lineorca513
educational, explorative, supportiveJust right4435Elimination of early tech issuesFun environment and lots of cool projects and ideasYesNo thankspython, R, bashORCA223
Educational, fun, inspiringJust right5555Have better abstracts and requirements, and also food to cater to all needs like the vegetarians. getting to play with new datasetsYesNo thanksshell, Ramazon server535
enjoyable, interesting, motivatingJust right4444Have clearer instructions for specific teams.Happy hour/dinnerYesNo thankssamtools, bamtools, python, Flask, BioDalliance, GATKAWS353
Exciting, enlightening, fun-filledJust right5453Have the project leads put more information prior to the start so more can be done during the three days of the hackathon.idea inspirationYesNo thanksR,ShinyR,Shiny111
exciting, grueling, enjoyableJust right4553Healthier food choices please!It was really awesome to see the collective depth of knowledge of everyone here at hackseq.YesNo thankspython,bash,html,css,js,samtools,bedtools,igv,biodalliance,picard,gatkVariant calling (lots of RAM and processors) and plenty of disk space354
fantastic,tiring,interesting,epic,fun,collaborativeJust right5554How to' guide for team leader (edited response)learning about new ideasYesNo thanksr shiny JavaScript Ogans laptop444
friendly,creative,deliciousJust right5555I really can't think of anything. The location worked perfectly, the tech stuff was all great. The coffee/food was as required.learning experience & collaborationYesNo thankspython 3ORCA513
fun, busy, collaborativeJust right5453I think it was perfect, maybe having a mic when talkinglearning from different peopleYesNo thankspython3ORCA54
Fun, collaborative, greatJust right5455I would eliminate the optional talks, and increase the time by a day or get teams to plan more beforehandLearnt a lot! Met some awesome people from varied backgrounds.YesNo thankspython, RAWS154can not think of right now
fun, frenetic, stimulatingJust right5555I would prune out projects that are not strongly selected by the participants to make sure everyone is very interested in the project they're assigned.Meeting and working with new people, learning from themYesNo thankspython, ipython notebook, bwa, samtools, longranger (10x genomics' software)AWS only, single large memory instance154
fun, impressed, workedJust right5555Internet access crapped out a couple times. Orcha was slow also.Meeting intelligent and friendly teammatesYesYesR, AWK, python, BASHAWS, ORCA355X11 forwarding and tunneling for RSTUDIO IPYTHON
fun, mindblowing, humblingJust right5555It was just great!Meeting new peopleYesYesR, pythonLaptop
fun, productive,informativeJust right5554It's pretty good already.Meeting new people and working as a teamYesYespythonORCA, but due to issues, many of us had to use our own clusters 514
fun, stressful, excitingJust right3333Maybe ensure that teams get evenly mannedMeeting new people in bioYesYesbash scriptsAWS343None.
Fun, teamwork, learningJust right4454Maybe some more intro-level material for people interested in bioinformatics but are new to it. Meeting new people with different skill sets and working on an interesting problemYesYespython,R,LongRanger,jyupiterJust AWS for data and jobs, and github for code. Created an HTTP server instance on AWS to host BAMs.155
fun,creavity,collaborativeJust right4445More information in terms of computing resources in advance of the hackathon, so that we don't waste time figuring that out at the event. Meeting other people and opening my mind to new ideasYesYesShell, python, heaps of toolsORCA was very slow to start with so a lot of us ended up using our home facility's HPC314
fun,hectic,challengingJust right5455More space to hack lateMeeting people, exploring technologiesYesYespythonORCA515nothing
funny, friendly, interestingJust right5454More teams. Option to combine teams at any pointMeeting so many people from other projects doing great amazing things! Sharung ideas and solutions too YesYesR, plotly and ShinyNothing.111Individual rooms for each team
good,nerdy,funJust right5555not overlap workshop with the teamwork, sometimes, one can not join workshop since the team is working togetherPeopleYesYesR, pythonORCA, AWS554
great, learning, team-buildingJust right3342Nothing! This was really perfect.Pizza on siteYesYespython, velvetAWS153A ready-set-go AWS instance & background material on the project
informative, fun, disorganizedJust right4554Overnight areas onsite Ribbing my teammatesYesYespython and RAWS355none
informative, fun, organizedJust right5554Prep by participants beforeSelected projects beforehand, no internet problems, excellent venue/food/organizationYesYesR,shiny,plotlyinternet and a laptop with R studio and git112none
informative, learning and wonderfulJust right5553Protein in the breakfasts!team work and learn from each other. YesYespythonORCA312I just wanted ORCA to work better
Informative,Amazing,CodejockeyJust right4353Put an address and a map on the website. Make sure information is communicated on all channels (e-mail, Slack, etc.). Don't overrun on the talks. Sort out all computational infrastructure.TeamworkYesYespythonAWS, 1 node153None
insightful, exciting, worth-itJust right5455Spread the news teamworkYesYesRJust laptop111none
inspiration, collaboration, awesomeJust right4334start later in the day, warmer space, more chances to get to know team in the beginningTesting out crazy new ideasYesYesR, pythonORCA334
inspiring,challenging,funJust right5554Students were passing through area during the last dayThe challengeYesYesBASH, python, Bedtools, Samtools, R, Junyper Notebooks (formerly ipython), GoogleDocsAWS was essentially for us as we needed over 4T of storage space, many nodes for processing data and a lot of RAM252
instructive,fun,intenseJust right5553The exact problem wasn't defined or explained very well beforehand. I had to spend the first day learning instead of coding.The collaboration and environment YesYespython,bash,ROur own clusters, orca333HPC
intense, enlightening, enrichedJust right5551The initial expectations within the team could be made more accessible - slack communication got a bit too chaotic to keep track of prior to the hackathon The collaborative aspect of delivering a project using different skill sets in the group.YesYesR, shinyour beloved laptops111
intense,dizzying,learningerificJust right5554Travel grants, extra time (outside of hacking) for events/workshopsThe novelty of projects and team compositions just perfect to handle to those situations. YesYespython3,PLINK1.9,IMPUTE2,bcftools1.3ORCA and a few heavy computational calculations on clusters at institutions back home.514Perhaps the ability to get a few hundred hours of CPU time on an easy to access system such as ORCA.
intense,productive,excitingJust right4455T-shirts?the opportunity to work with an expert and collaborate to resolve a relevant scientific problemYesYespython, R, pysamAWS. Fairly light compute requirements in our case.5Biggest IT issue: no open ports to AWS instance -- made it hard to get ad-hoc servers running (ipython notebook, rstudio, http mirroring, etc). Per-user accounts on AWS: needed to spend a bunch of time configuring these. (may just be my lack of experience w/ AWS), would be good to have in HOWTO?
intense; unexpected; funJust right5555Unsure if people with my level of inexperience have much of a place in teamsThe team bondingYesYespython (conda, pyasm etc.), ipython, bash, make, longranger (10x genomics), RAWS, various servers on the machine.154More time to establish accounts on the machine prior to the start. We were all using same user account and it could possibly lead to a disaster.
interactive, friendly,rewardingJust right5555We could probably have more information about the project before hand so we are able to select our preferences betterThe wonderful people!YesYesR package, python55
interesting, intense, funJust right5555Three days was a great amount of time - wonderful to meet with people. YesYES!bash,pytho,webAWS154
Interesting,conceptual,algorithmicJust right4354VancouverYesYES!python3,plink,R,vcftools,bcftools,impute2ORCA and cluster at one of the participants institutions414
Interesting,educational,coolJust right5545variety of projectsYesYES!R, python, bash, memesuiteDidn't use AWS that much, it was mostly run on our laptops335None
interesting,great,fantasticJust right5555Working as a team, brainstorming ideas, full day sessions to focus on work, Meeting deadlineYesYES!python,Ipython notebook,bwa,velvetUsed a single node in AWS. Worked brilliantly155Really nothing. The 10x guys got basic bioinformatics tools set up on the AWS node very quickly. Might have been useful for other teams to make that easier but we had absolutely no problems
Learning, inclusive, productiveJust right3433Working with a new set of people, the wrap ups at the end of the day and the venue. YesYES!bash, python, R, bedtools, ChIPSeeker, MotifGPWe didn't use AWS or ORCA. It wasn't clear how to access ORCA, and AWS required some set-up that we didn't want to go through. I would recommend that if you're going to use AWS in a future hackathon, then have the instances ready to go so that the participants don't have to worry about configuring the servers.115Preconfigured cloud instances.
productive,ambitious,exhaustingToo Long4354YesYES!bashAWS, ORCA555NA
supercalifragilisticexpialidociousToo Long4443YesYES!python, nextflow, bashWe used ORCA, and resources from our home intitutions325absolutely nothing!
tantalizing, intensive, rewardingToo Long5555YesYES!python, R; Optimization packages & blackbox implementationsBasic programming capabilities, CLI, Version control (Git/Github), experience with packages & libraries434N/A
Too Short4454YesYES!python,bash,magicblastAWS,ORCA554
Too Short4455YesYES!python, BashAWS servers and ORCA351None!
Too Short5553YesYES!pythonAWS 16 cores, 128 GB RAM341The resource is great, no need to ask for more
Dataset 2.Post-hackseq survey responses.
De-identified post-hackseq survey response data for the figures.

Technical and logistical requisites

Hackathons have little essential resource requirements. In this section, we outline the core logistics and technical infrastructure we employed. While these requisites could be stripped down, our experience was that attention and planning for these details maximized the efficiency of our teams to focus on coding and development.

Core logistics

To encourage participation, hackseq had no entry-cost for participants. To ensure teams could focus on the hackathon and not technical or logistical issues, we secured funding for the venue, technical infrastructure, food, transportation and stationery by partnering with different organizations.

A sponsorship package was created to approach different academic, non -profit and industry organizations. Besides asking for financial support, we also made communication and marketing requests given that one of hackseq’s goals was to recruit a diverse pool of participants. A strong emphasis was placed on women’s groups in science and technology.

In November 2015, we contacted ASHG to ask if we could be a satellite event for their meeting. Given that ASHG 2016 conference was planned to be held in Vancouver, hackseq gained exposure from the ASHG's communication strategy. The ASHG also provided three travel grants to participants based on financial need and diversity.

These partnerships allowed hackseq to take place in a large, bright atrium at the University of British Columbia. This allowed all the teams to be in a single-space and interact with one another. Food was provided to minimize distraction and two social events were hosted, one the first night and one on the last night to encourage collaboration and networking amongst participants.

Technical infrastructure

Reliable technical infrastructure is necessary for organizing a successful hackathon; primarily, electrical power, Internet access and computing resources. We ensured the venue had adequate electrical outlets for the participants’ laptops and organized a dedicated Wi-Fi network connection be established for the event through the university's information technology office.

Unlike many hackathons, hackseq was not restricted to coding. It also included genomic data analyses, which required additional computational resources. To promote reproducibility and collaboration, all the projects were based on pre-organized GitHub teams and repositories (see hackseq organization on Github; https://github.com/hackseq). To provide teams with reliable and powerful computation, we secured in-kind donation of cloud computing from Amazon Web Services Elastic Compute Cloud (AWS-EC2), and Canada’s Michael Smith Genome Science Centre genOmics Research Container Architecture (ORCA). We used Linuxbrew, a cross-platform package management tool, to install bioinformatics software on ORCA5.

There was an equal usage of AWS-EC2 and ORCA amongst the participants (43%, not mutually exclusive) and an additional 12% using high-performance computing resources from their resident institutions. Users showed a preference for resources they were previously experienced with, and reported that it was not feasible to learn to use a new computing resource in the given time. Allowing team leaders and participants access to computing resources ahead of time in the future to ‘experiment’ and familiarize themselves with the different resources is advisable.

Each team chose which programming language and software they used. The majority of participants relied on Python (82.6%) and R (53.8%) programming languages and also used specialized software that related to their particular projects (Figure 2).

b5411d73-9bbb-41da-93a3-8a6ebff9b3c4_figure2.gif

Figure 2. Software usage during hackseq 16.

At the conclusion of hackseq, we asked participants to complete a survey on their experience at hackseq. There were 52 responses to the question, “Which programming languages and tools did you and your team use during the course of hackseq? (Comma delimited please).” These responses were parsed and the number of unique instances is reported. Languages or software listed <2 is reported as ‘Other’.

In summary, the infrastructure requisites for running a successful hackathon are minimal and many can be acquired as in-kind donations from related organizations. In highlighting the essentials and key lessons, we hope to encourage the motivated reader to run a local scientific hackathon.

Research project summaries

The projects undertaken during hackseq were from diverse fields within bioinformatics, ranging from human genomic variation analysis, microbial ecology and transcriptomics, to bioinformatic algorithm development. The projects were proposed by the team leaders, who defined the scope of the work, with the idea that at the end of the 72 hours there will have developed a working prototype. At hackseq, the teams organized organically, with team leaders defining the problem and teaching the participants the necessary background while participants offered their expertise on how to implement a solution. In this way a collaborative project goal could be explicitly defined (and re-defined), and the teams would work towards completing that goal together.

Here we provide brief summaries from the projects. Scientific abstracts, videos of final presentations and updated information on each project can be found at www.hackseq.com/projects16.

VASCO: Visualization app for single cell exploration (led by Grace X.Y. Zheng)

Modern transcriptomics analysis tools have limited capacity for analyzing thousands of single-cell RNA-sequencing data (scRNA-seq). VASCO is an intuitive user-interface to visualize gene-cell expression and cell clustering data to explore the relationship between populations of cells and gene expression, including cell cluster of differentiation markers (CD-markers). This project was awarded the “People’s Choice” for the most outstanding project developed at hackseq.

XYalign: Hacking sex chromosome variation (led by Melissa A. Wilson Sayres)

Human sex chromosomes violate typical ploidy assumptions made for NGS autosome copy number and variant measurement, which is further confounded by mis-alignment between the X and Y chromosomes. XYalign was developed to measure sex chromosome ploidy and remap reads based on the inferred sex for downstream analysis.

ParetoParrot: A tool to optimize the parameters of command line software (led by Shaun Jackman)

Many bioinformatics software, such as genome sequence alignment and assembly, requires optimization of several input parameters to maximize a target metric. ParetoParrot measured the performance of several ‘black-box’ optimization algorithms to improve the performance of genome sequence assembly software.

BaklavaWGS: Pseudo-WGS variant calling for common cell types aggregating ChIP-seq, RNA-seq and DHS from ENCODE and Roadmap Epigenomics data (led by Luca Pinello)

There is a wealth of sequencing datasets for cell types that have helped to understand and prioritize non-coding variants. Unfortunately, for many of those cell types we still don't have complete genotype information. BaklavaWGS recovers genotype data from cell lines aggregating sequencing data to aid downstream allele specific analysis. A preliminary analysis is available at http://www.baklavawgs.com/.

Evaluating epigenetic modifications in ChIP-seq and methylation data across cell types and states (led by Manuel Belmadani)

A variety of datasets and approaches were investigated for analyzing cell type and state-specific genome regulation. The outcome of the experimental work in exploring differentially methylated regions from different epigenomic data and public databases, such as ENCODE ChIP-seq, IHEC and JASPAR, is presented.

Selection of tag SNPs for an African SNP array by LD and haplotype based methods (led by Tommy Cartensen)

Commercial SNP arrays fail to capture the diversity of African populations and limit the capacity to conduct large-scale medical genetic studies. Using African whole genome sequencing (WGS) data, an algorithm was developed to quickly identify SNP tags for this population. This will be used to improve upon SNP arrays for this richly diverse continent.

Somatic mutation from separated haplotypes (SMUSH) (led by Patrick Marks)

Calling somatic mutations relies on matched tumour and normal DNA sequencing, but a matched normal sample is often not available. The SMUSH algorithm was developed to differentiate wild type, germline and somatic mutations from linked-read DNA sequencing libraries.

MetaGenius (led by Michael Schnall-Levin)

Analysis of shotgun metagenomic sequencing data is limited in its capacity to assemble over homologous sequences. MetaGenius uses linked-read DNA sequencing to improve the assembly of a mixture of five bacterial species.

mICP: Metagenomic indicator contig predictor (led by Ben Busby)

Metagenomic sequencing has largely focused on 16S rRNA amplicons. This mICP strategy uses a mixture of long PacBio and short Illumina reads to identify contigs from environmental sequencing samples, which predict the environmental state from which they were found.

Discussion

The overarching themes of hackseq were inclusivity, open science and collaboration. To gauge the extent to which we were successful in delivering on these themes, we performed a final survey at the conclusion of hackseq. Participants overwhelmingly described their experience as positive (Figure 3), with 100% (n = 54) of the survey respondents indicating that they would participate in an event like hackseq again and a further 80% indicating that they would like to take on an organizational/leadership role in future hackseq events. Participants specifically highlighted that hackseq created ample recruitment, employment and collaborative opportunities, while also exposing participants to different datasets and analyses. We believe this reflects the underlying desire amongst young scientists to share, collaborate and learn from one another. They only need be given the opportunity to do so.

b5411d73-9bbb-41da-93a3-8a6ebff9b3c4_figure3.gif

Figure 3. Quantification of subjective experience.

To measure the quality of the experience hackseq participants had after the event, we asked (A) “Please write three single word adjectives to describe your experience at hackseq?” Responses were parsed and used to make a word-cloud (www.wordle.net), where the size of the word is proportional to the number of occurrences of that word in the survey responses. For scale, in 50 responses: ‘fun’ was mentioned 26 times; ‘exciting’ 6 times; and ‘supercalifragilisticexpialidocious’ once. (B) Additionally, we asked participants to rate four dimensions of their experience on a linear scale from 1 to 5. The kernel density of responses for these dimensions are shown, with a red dotted line showing the mean value of the responses.

By organizing hackseq as a satellite meeting of an international conference like ASHG, we were able to attract team leaders and participants from around the world, including a large proportion of young investigators and female participants (Figure 1). There was a higher proportion of females at hackseq (35.5%), than reported ratios at hackathons for which data is available, 20% at NASA’s Space Apps Challenge (https://www.fastcompany.com/3059036/most-creative-people/what-do-women-want-at-hackathons-nasa-has-a-list) or 15% at Spotify-organized hackathons (https://labs.spotify.com/2015/01/13/diversify-how-we-created-a-hackathon-with-50-50-female-male-participants/), which we believe to be a consequence of starting with a representative organizing committee and specifically encouraging female participation during recruitment. Although, this comparison is confounded by differences in starting demographics between computer science/engineering students and bioinformatics/biology students.

To further increase global representation at future hackseq events, we recommend providing additional targeted travel awards or remote participation options to reduce proximity/cost restrictions. Further improvements could include educational resources to address common technical issues, the provision of an overnight area for participants who would like to continue to work after hours and additional activities to encourage interaction with members from different teams.

Conclusion

The nature of biological sciences has shifted to an increasing emphasis on computational analysis. Collaborative events, such as hackseq, offer an exciting platform to bring together a wide spectrum of scientists to work together and innovate. We present demographic information about the first hackseq hackathon and encourage future organizers to do likewise, to quantify social inequalities that may be present in such events, and strive to achieve equal representation in the sciences. It’s our hope that the information presented here will aid and encourage others in organizing genomics hackathons.

Data availability

Dataset 1: hackseq demographics: De-identified demographic data from hackseq participants in the pre-meeting survey/confirmation of attendance. doi, 10.5256/f1000research.10964.d1528026

Dataset 2: Post-hackseq survey responses: De-identified post-hackseq survey response data for the figures. doi, 10.5256/f1000research.10964.d1528037

Comments on this article Comments (0)

Version 2
VERSION 2 PUBLISHED 28 Feb 2017
Comment
Competing interests
Grant information
Copyright
Download
 
Export To
metrics
Views Downloads
F1000Research - -
PubMed Central
Data from PMC are received and updated monthly.
- -
Citations
CITE
how to cite this article
hackseq Organizing Committee 2016. hackseq: Catalyzing collaboration between biological and computational scientists via hackathon [version 2; peer review: 2 approved]. F1000Research 2017, 6:197 (https://doi.org/10.12688/f1000research.10964.2)
NOTE: If applicable, it is important to ensure the information in square brackets after the title is included in all citations of this article.
track
receive updates on this article
Track an article to receive email alerts on any updates to this article.

Open Peer Review

Current Reviewer Status: ?
Key to Reviewer Statuses VIEW
ApprovedThe paper is scientifically sound in its current form and only minor, if any, improvements are suggested
Approved with reservations A number of small changes, sometimes more significant revisions are required to address specific details and improve the papers academic merit.
Not approvedFundamental flaws in the paper seriously undermine the findings and conclusions
Version 2
VERSION 2
PUBLISHED 10 Apr 2017
Revised
Views
8
Cite
Reviewer Report 08 May 2017
Kate L. Hertweck, Department of Biology, The University of Texas at Tyler, Tyler, TX, USA 
Approved
VIEWS 8
My comments ... Continue reading
CITE
CITE
HOW TO CITE THIS REPORT
Hertweck KL. Reviewer Report For: hackseq: Catalyzing collaboration between biological and computational scientists via hackathon [version 2; peer review: 2 approved]. F1000Research 2017, 6:197 (https://doi.org/10.5256/f1000research.12239.r21700)
NOTE: it is important to ensure the information in square brackets after the title is included in all citations of this article.
Views
6
Cite
Reviewer Report 13 Apr 2017
Jiarong Guo, Center for Microbial Ecology, Michigan State University, East Lansing, MI, USA 
Approved
VIEWS 6
My previous comments have been addressed by authors in the new version ... Continue reading
CITE
CITE
HOW TO CITE THIS REPORT
Guo J. Reviewer Report For: hackseq: Catalyzing collaboration between biological and computational scientists via hackathon [version 2; peer review: 2 approved]. F1000Research 2017, 6:197 (https://doi.org/10.5256/f1000research.12239.r21699)
NOTE: it is important to ensure the information in square brackets after the title is included in all citations of this article.
Version 1
VERSION 1
PUBLISHED 28 Feb 2017
Views
11
Cite
Reviewer Report 31 Mar 2017
Jiarong Guo, Center for Microbial Ecology, Michigan State University, East Lansing, MI, USA 
Approved
VIEWS 11
The authors reported a detailed summary of their genomic hackathon, hackseq, to help those interested in organizing similar hackathons in future. The hackseq brought together 66 biological and computational scientists with diverse demographic background to collaborate on nine projects on ... Continue reading
CITE
CITE
HOW TO CITE THIS REPORT
Guo J. Reviewer Report For: hackseq: Catalyzing collaboration between biological and computational scientists via hackathon [version 2; peer review: 2 approved]. F1000Research 2017, 6:197 (https://doi.org/10.5256/f1000research.11818.r20964)
NOTE: it is important to ensure the information in square brackets after the title is included in all citations of this article.
  • Reader Comment 03 Apr 2017
    Shaun Jackman, BC Cancer Agency Genome Sciences Centre, Canada
    03 Apr 2017
    Reader Comment
    Thank you for your review, Jiarong.

    > The team leaders seem to have a critical role in each project, but their roles and responsibilities during the hackathon are not clearly mentioned ... Continue reading
COMMENTS ON THIS REPORT
  • Reader Comment 03 Apr 2017
    Shaun Jackman, BC Cancer Agency Genome Sciences Centre, Canada
    03 Apr 2017
    Reader Comment
    Thank you for your review, Jiarong.

    > The team leaders seem to have a critical role in each project, but their roles and responsibilities during the hackathon are not clearly mentioned ... Continue reading
Views
22
Cite
Reviewer Report 31 Mar 2017
Kate L. Hertweck, Department of Biology, The University of Texas at Tyler, Tyler, TX, USA 
Approved
VIEWS 22
Thank you to the authors for writing a summary of what seems to be a very successful collaborative coding event, with this manuscript in particular focused on preparation for the event, managing logistic concerns during the event, and an overview ... Continue reading
CITE
CITE
HOW TO CITE THIS REPORT
Hertweck KL. Reviewer Report For: hackseq: Catalyzing collaboration between biological and computational scientists via hackathon [version 2; peer review: 2 approved]. F1000Research 2017, 6:197 (https://doi.org/10.5256/f1000research.11818.r20626)
NOTE: it is important to ensure the information in square brackets after the title is included in all citations of this article.

Comments on this article Comments (0)

Version 2
VERSION 2 PUBLISHED 28 Feb 2017
Comment
Alongside their report, reviewers assign a status to the article:
Approved - the paper is scientifically sound in its current form and only minor, if any, improvements are suggested
Approved with reservations - A number of small changes, sometimes more significant revisions are required to address specific details and improve the papers academic merit.
Not approved - fundamental flaws in the paper seriously undermine the findings and conclusions
Sign In
If you've forgotten your password, please enter your email address below and we'll send you instructions on how to reset your password.

The email address should be the one you originally registered with F1000.

Email address not valid, please try again

You registered with F1000 via Google, so we cannot reset your password.

To sign in, please click here.

If you still need help with your Google account password, please click here.

You registered with F1000 via Facebook, so we cannot reset your password.

To sign in, please click here.

If you still need help with your Facebook account password, please click here.

Code not correct, please try again
Email us for further assistance.
Server error, please try again.