This month’s “Company Spotlight” takes a closer look at Othram Inc which applies the power of modern sequencing technology and genomics to forensics. David Mittelman, Chief Executive Officer of Othram, discussed with us in detail Othram’s approach towards developing high quality sequencing methodology and building software to help solve cold cases with forensic DNA analysis.
Othram develops sequencing protocols and software required to reconstruct genomes from degraded and low volume DNA samples. The recently (2018) launched ten-person company is headquartered in The Woodlands, TX, and founded by David Mittelman (previously co-founder/CSO of Arpeggi and CSO at consumer genetics firm Family Tree DNA) and Steven Hsu who adds AI & machine learning expertise to the team.
The following summarizes questions and answers from my dialogue with David Mittelman, CEO and co-founder of Othram.
EB: Tell us about Othram – Your business is focused on delivering reconstructed genome sequences from degraded and low-input DNA sources. What need(s) are you trying to address and what products/services do you offer?
David Mittelman: When a lab starts with fresh DNA samples (e.g. saliva samples sent to consumer genetics companies) it is generally straightforward to perform genetic analysis. There are other circumstances when a lab doesn’t have good quality, high quantity DNA to start with. For example, when a medical lab starts with circulating cell-free DNA, which can be released from tumor cells slipping into the blood stream, it is usually degraded and sparsely available. There are similar scenarios in which fetal DNA is isolated from the bloodstream of mothers. In forensics, there are a lots of cold cases, unidentified remains, historical projects, and ancient DNA projects – most of which involve very small amounts of aged material that yield little to no DNA. Even when there is DNA, it’s damaged and degraded.
Othram combines unique laboratory processes with our own software algorithms, to identify and eliminate noise so that we can better identify genetic data that truly corresponds to a DNA sample. We use this data to help our customers learn more about the identity of unknown persons and the circumstances in which their remains were found.
Othram is focusing with its services on criminal justice cases. How did this come about?
DM: Our team at Othram consists of veterans in the genomics field. Some of us have a couple decades of experience working with genetic data. We really want to apply genomics to an important problem with clear public benefit and that others are not working on. The forensic space is an underserved market for genomics, and it is really easy to bring value to the community. For every case solved, there are lots of suspects exonerated and there is closure for families. In the case of unidentified remains, we are able to give a voice to the nameless and re-anchor them into the community.
EB: Can you provide some specific examples of how your services are used / applied.
DM: Right now we are focused on cold cases involving unidentified remains. We work with local law enforcement to help them test DNA from these remains. Right now, crime DNA testing is principally based on the National DNA Index System (NDIS). You can search the FBI database using CODIS markers for previously established identities. CODIS is the acronym for the Combined DNA Index System. Originally composed of 13 markers, the set of markers has been expanded to 20 markers. These markers can be used to confirm someone’s identity if they are in the database, but CODIS markers cannot be used to find someone new and they have limited utility in relationship testing.
Our DNA tests complement CODIS marker-based tests. If a conventional test with CODIS markers fails to confirm a known identity (common for victims, children, and perpetrators that have yet to be caught), we can perform genome-wide sequencing to determine ancestry, identify quantitative physical traits, and find nearest relatives. Sometimes, as with the now famous case of the Golden State Killer, it is possible to establish an identity through the triangulation of genealogical records.
EB: Where do you see Othram’s strength in reconstructing genome sequences, on the sequencing side (i.e. sequencing challenging samples), or on the computational side (applying internally developed and optimized algorithms) or both?
DM: To properly address the challenge of sequencing degraded samples, you need to have expertise in both and this actually ties well into my background and expertise acquired while working both in the private and public sectors: helping build the STR Variant Caller for the 1,000 Genomes Project, building a suite of software tools at Arpeggi Inc. (acquired by Gene-by-Gene) to help people get useful information from Illumina sequencers, and benchmarking NGS pipelines via the Genome in a Bottle consortium hosted by NIST. At Othram, my team and I are taking the same tools and learned lessons, and we are translating them for use in forensics. We are doing a lot of software development from the ground up and we are fine tuning and validating existing tools. Our R&D team performs experiments in the lab with known samples, known variants, and known mixtures of samples and then validates our software methods to verify that we are able to arrive at the correct answer. Like with previous companies we’ve been involved in, we will publish all the work to drive the field forward. We want to bring to forensic sequencing what we have done for medical sequencing in the past: tools for standardizing and performance testing DNA sequencing and other forms of DNA testing. This is a top priority for us and since most of us are former academic researchers, we are eager to publish and share our finding with the greater scientific community.
Othram applies machine learning models for prediction purpose. Can you provide more details?
DM: My co-founder, Steve Hsu is a machine learning expert. He builds machine learning algorithms for building polygenic scores. He has built models for a number of physical traits, like height, while at Michigan State University. Steve published some work that shows that he can predict height down to inches. So yes, that is an ongoing effort on our part to use tens of thousands of genetic changes to accurately identify fine-grained variation in physical traits.
Who does Othram cater to or target with its products / services? How are your services superior over existing approaches (e.g. marker identification).
DM: Our main target market is law enforcement, such as state and local government agencies. We also work with crime labs, both public and private. There are even non-forensic applications of our technology that we offer to commercial companies outside of forensics. After our SXSW 2019 competition win we got approached by some of them and I guess there is word of mouth and emerging general awareness about our activities and products.
This sector poses a different challenge from what we saw before in the healthcare sector. There are not a lot of players in forensic genomics so we do spend a lot of time on education and validation to win the trust of customers and stakeholders. Sometimes I feel like it’s the early days of the WWW and we are the AOL of forensics.
Othram recently raised $4 million in Series A. How do you intend to use these funds?
DM: We raised $4M from Nimble Ventures in November 2018 to build a state-of-the-art forensic genomics laboratory. We also won the SXSW Illumina startup pitch competition in March of this year which yielded a $10,000 Illumina research grant to explore forensic sequencing on Illumina platforms.
Our first priority was to build out our sequencing laboratory which we completed now. It’s an almost 4,000 square foot lab with three sequencers so far: a one NovaSeq, a MiSeq, and an iSeq. We also have other hardware for sample preparation and QC. Importantly, we build a forensics lab that houses a sequencing operation. This is very different from a genomics lab that sequences forensic samples. There are lots of infrastructure, procedural, and talent requirements for operating a forensics lab. To complement the laboratory, we are also building out a software team.
You mention the team at Othram currently consists of 10 people. How is the company divided in terms of sequencing laboratory versus the computational analysis side – how many work in the lab versus how many are developing optimized analysis software?
DM: Currently, we have three individuals working in the lab and seven on the software side. I split myself half laboratory and half software development.
What are some of the biggest challenges you have to overcome when sequencing degraded, low-quality samples? How do you go about optimizing your laboratory protocols to overcome these challenges?
DM: There are three major problems we face when sequencing forensic DNA samples:
- Contamination: A lot of the samples we receive are contaminated with bacterial DNA and non-human DNA. At times it is hard to efficiently sequence the human samples and as such if you amplify the human DNA samples you also amplify non-human DNA. We have developed ways to deplete or enrich certain portions of a sample and we are working to improve and scale those methods.
- Not enough good quality DNA: If the DNA has been treated poorly or is very old, it begins to degrade. A lot of the sample material received consist of DNA fragments of too small a size which are challenging to use in a reaction. We are always trying to push the lower limit of what we can accept for sequencing.
- Mixtures: We sometimes process samples composed of more than one person’s genome and we are working on tools to help de-convolute these mixtures.
Who do you view as your current competition and why? What differentiates Othram from other players in the market?
DM: As I mentioned earlier, this is not currently a very crowded market and as such we are not facing a lot of competition. There are a few companies that are trying to offer support in the area of forensics, but they don’t have a lab. One example is the DNA Doe Project, an initiative that uses genetic genealogy to identify John and Jane Does, an organization that works for law enforcement agencies and medical examiners across the country to help solve intractable cases. Since they don’t currently have a lab, we support them on the sequencing side where our expertise lies and where we are invested in. They have a lot of samples that need to be processed. They work nationally with law enforcement. Until recently they were doing SNP arrays and now they are doing sequencing and we support them in those efforts. So really we have few competitors and instead, lots of collaborators and partners.
What do you see as the biggest challenge(s) the genomics data field is currently facing and why? How can we overcome these challenges? How will this affect what Othram does?
DM: One of the most interesting and also important challenges is the implementation of validated whole genome sequencing for many types of samples. To do this right, it needs to be standardized which requires rigorous performance testing, no matter whether it is in the direct-to-consumer, medical, or forensic setting. This requires the implementation of an optimized and validated sequencing lab which generates information that is useful – as the saying goes “garbage in, garbage out”. To implement such a process one needs to fully understand the wet lab components and how to optimize, validate, and test them to get meaningful answers. You also need software to monitor the lab process and report deviations and failed controls. There are many factors to evaluate and understand which includes acceptable contamination levels, proportions of mixtures, and minimum quantities and qualities of DNA.