I’m new to this mailing list, and to DHIS2 in general, so would like to start by saying hi to everyone!
I’m looking to create an DHIS2-based app that displays viral sequencing information on top of clinical cases (something along the lines of NextStrain). Is this possible? Would I need to modify the underlying DataElement in dhis2-core to support genetic metadata? Or would it be enough to add a reference to sequences stored elsewhere, for example in GenBank?
Welcome to the list! From what I understand, you want to associate a given strain, such as MK13013 or the longer version Ebola virus/H.sapiens-wt/SLE/2015/Makona-1309_C2_MK13013, with a clinical case? That should be possible in several ways, either as a free text data element, an option set, or as metadata attribute
If you want more of the meta-data such as the complete sequence, it should in theory be possible, but a link to the ncbi would perhaps be the best.
Regards,
Johan Ivar Sæbø
···
-----Original Message-----
From: Dhis2-users [mailto:dhis2-users-
bounces+johansa=ifi.uio.no@lists.launchpad.net] On Behalf Of Andres
Colubri
Sent: Thursday, April 06, 2017 3:40 PM
To: dhis2-users@lists.launchpad.net
Subject: [Dhis2-users] storing/linking sequence metadata in DHIS2
I'm new to this mailing list, and to DHIS2 in general, so would like to start by
saying hi to everyone!
I'm looking to create an DHIS2-based app that displays viral sequencing
information on top of clinical cases (something along the lines of NextStrain
<auspice; ). Is this possible?
Would I need to modify the underlying DataElement in dhis2-core to support
genetic metadata? Or would it be enough to add a reference to sequences
stored elsewhere, for example in GenBank?
Many thanks for your kind reply. Yes, that’s correct, we would like to associate pathogen sequence data to clinical cases. It sounds like storing the link to the NCBI records in a metadata attribute would be the simplest solution, as long as the sequences are already deposited publicly somewhere else. Then the app would pull the links from the DHIS2 DB and retrieve the full sequences. We will look into it, and will get back with more questions.
Best,
Andres
···
On Fri, Apr 7, 2017 at 4:00 AM, Johan Ivar Sæbø johansa@ifi.uio.no wrote:
Would I need to modify the underlying DataElement in dhis2-core to support
genetic metadata? Or would it be enough to add a reference to sequences
stored elsewhere, for example in GenBank?
Best,
Andres
Hi Andres,
Welcome to the list! From what I understand, you want to associate a given strain, such as MK13013 or the longer version Ebola virus/H.sapiens-wt/SLE/2015/Makona-1309_C2_MK13013, with a clinical case? That should be possible in several ways, either as a free text data element, an option set, or as metadata attribute
is seems to be a bunch of flu strains. I guess something similar exists for Ebola?
The approach we have taken with another system is to transform and import these into DHIS2 through an ETL process. I think something similar might be possible. It might be worth looking into and I would be happy to discuss further with you about our approach.
Many thanks for your kind reply. Yes, that’s correct, we would like to associate pathogen sequence data to clinical cases. It sounds like storing the link to the NCBI records in a metadata attribute would be the simplest solution, as long as the sequences are already deposited publicly somewhere else. Then the app would pull the links from the DHIS2 DB and retrieve the full sequences. We will look into it, and will get back with more questions.
On Fri, Apr 7, 2017 at 4:00 AM, Johan Ivar Sæbø johansa@ifi.uio.no wrote:
Would I need to modify the underlying DataElement in dhis2-core to support
genetic metadata? Or would it be enough to add a reference to sequences
stored elsewhere, for example in GenBank?
Best,
Andres
Hi Andres,
Welcome to the list! From what I understand, you want to associate a given strain, such as MK13013 or the longer version Ebola virus/H.sapiens-wt/SLE/2015/Makona-1309_C2_MK13013, with a clinical case? That should be possible in several ways, either as a free text data element, an option set, or as metadata attribute
is seems to be a bunch of flu strains. I guess something similar exists for Ebola?
The approach we have taken with another system is to transform and import these into DHIS2 through an ETL process. I think something similar might be possible. It might be worth looking into and I would be happy to discuss further with you about our approach.
Many thanks for your kind reply. Yes, that’s correct, we would like to associate pathogen sequence data to clinical cases. It sounds like storing the link to the NCBI records in a metadata attribute would be the simplest solution, as long as the sequences are already deposited publicly somewhere else. Then the app would pull the links from the DHIS2 DB and retrieve the full sequences. We will look into it, and will get back with more questions.
On Fri, Apr 7, 2017 at 4:00 AM, Johan Ivar Sæbø johansa@ifi.uio.no wrote:
Would I need to modify the underlying DataElement in dhis2-core to support
genetic metadata? Or would it be enough to add a reference to sequences
stored elsewhere, for example in GenBank?
Best,
Andres
Hi Andres,
Welcome to the list! From what I understand, you want to associate a given strain, such as MK13013 or the longer version Ebola virus/H.sapiens-wt/SLE/2015/Makona-1309_C2_MK13013, with a clinical case? That should be possible in several ways, either as a free text data element, an option set, or as metadata attribute
As Johan mentioned, the best way to do this probably would be with an option set. Looking at NextStrains process, they import some metadata on Ebola strains (https://github.com/nextstrain/augur/blob/master/ebola/ebola.py#L117) and process this into the format they need for the website. I think it should be possible to do the same for DHIS2, and process this into an option set (https://docs.dhis2.org/master/en/developer/html/webapi_csv_metadata_import.html#webapi_csv_option_sets) which could then be imported into your DHIS2 database. Once you have the metadata in the correct format, it would just be a matter of importing it as an option set into DHIS2. Once you do that, you could then use the events module to link an event (like a case detection) .
The link you provided looks much more unstructured, so it might be more difficult to parse and scrape this data.
is seems to be a bunch of flu strains. I guess something similar exists for Ebola?
The approach we have taken with another system is to transform and import these into DHIS2 through an ETL process. I think something similar might be possible. It might be worth looking into and I would be happy to discuss further with you about our approach.
Many thanks for your kind reply. Yes, that’s correct, we would like to associate pathogen sequence data to clinical cases. It sounds like storing the link to the NCBI records in a metadata attribute would be the simplest solution, as long as the sequences are already deposited publicly somewhere else. Then the app would pull the links from the DHIS2 DB and retrieve the full sequences. We will look into it, and will get back with more questions.
On Fri, Apr 7, 2017 at 4:00 AM, Johan Ivar Sæbø johansa@ifi.uio.no wrote:
Would I need to modify the underlying DataElement in dhis2-core to support
genetic metadata? Or would it be enough to add a reference to sequences
stored elsewhere, for example in GenBank?
Best,
Andres
Hi Andres,
Welcome to the list! From what I understand, you want to associate a given strain, such as MK13013 or the longer version Ebola virus/H.sapiens-wt/SLE/2015/Makona-1309_C2_MK13013, with a clinical case? That should be possible in several ways, either as a free text data element, an option set, or as metadata attribute