oyvble / euroformix Goto Github PK

There is an issue regarding the use of marker specific settings.
The issue happens when doing following:

1. Save marker specific settings for one particular dataset
2. Load another type of dataset with other markers than in the first dataset.
3. An error will occur when evaluating the new dataset, and marker specific settings are not set again: The function setVecRightOrder throws the error "The setting vector was not same length as the number of markers (mismatch)"

A quick fix is to save the Marker specific settings again for the new dataset.

I am missing an error message below:

library(euroformix, quietly = T)
freq = list("L1" = c("1" = 0.9, "2" = 1.1))
genDataset(2, freq, mx = c(0.8, 0.2))[[2]][[1]]
#> $L1
#> $L1$adata
#> [1] "2"
#> 
#> $L1$hdata
#> [1] 1965

^{Created on 2019-12-16 by the reprex package (v0.3.0)}

The function prepareC function (L244-L245) contained a bug causing problems for the stutter-model(s) - using BW, FW or both - when the Q-allelel is not defined.

• This only happens when the observed alleles in the evidence profile fully explains the the allele frequency outcome

Hi!

I noticed the repo has no licence associated. I believe the project is intended to be open-source, and therefore this might be an issue.

According to this thread on StackOverflow the software can not be used unless a private licence is granted to the users. To fix this, an open source licence should be added to the repo. Given the lack of a license I am actually not allowed modify the code to add a licence and create a pull request :). Therefore I would kindly ask you to add one. For a review of available options please see https://choosealicense.com/

Greetings!

I am running EuroForMix 2.2.1 using the ENFSI data posted on the website. I have no problem importing the data and configuring the model, but when I attempt to run the Quantitative LR I get the following message on the console.

Error in function(x) dbeta(x, 1, 1) : could not find function "function"

The dbeta(x, 1, 1) looks like a parameter for the stutter model. I have the stutter model turned off, so I'm wondering why this message is popping up. Please advise.

In case it helps, I'm running R 3.6.1, and I was able to reproduce this error on Mac OS X "Mojave" as well as Ubuntu 18.04. Thanks!

A crash happens when trying to put a person of interest (POI) under Hd instead of Hp. The reason for this is how the POI is extract after calculations in order to calculate the "Upper boundary LR".

Thanks to Damir Tesanovic for discovering this issue.

A crash happens using the GUI when following these steps:

1. Load an old EFM project (created with a version before v3.0.0)
2. Save “settings”.
3. Restart EFM and try set “settings” again.

Every time I click on "View Evidence" or want to view any graph, after the graphs are opened, the program is frozen and I cannot look the complete graphs.
On my Mac terminal I always get this error message:

Any idea on how to fix it?

Following error occur when clicking "Import from STRidER" in the EuroForMix GUI (v3.3.0):

Error in `[.data.frame`(value, , 1, drop = TRUE) : 
  undefined columns selected

A bug has been found regarding saving marker specific settings in the GUI.
Current version is unable to properly read marker names if they contain spaces (e.g. Fusion 6C).
As a temporary fix, users can save a project after saving the settings, and open project again after closing the GUI.

When testing an evidence sample that is degraded, one database completes successfully (African American), while another does not (Asian) when the degradation box is checked "YES" in the GUI. I have tested the attached evidence sample in both the current EFM GUI, as well as an Rmd script that I use, and get the same results in both, for both databases. The gui returns the error "The specified model could not explain the data" for the Asian database, while the AfAm database returns a log10LR in the 20's. The databases are from the 2017 NIST revised frequencies.

The aberrant behavior can be traced to the contLikMLE function when condOrder = hpcond. For the Asian database, the degradation slope cannot be computed, and all loci return NA. The hdcond degradation slope computes properly. When tested with two other ethnic dbs (Caucasian and Hispanic, also from the NIST 1036 study), proper LRs are computed with the degradation option on. This behavior does NOT occur when the degradation box is checked "NO" in the GUI; the MLE completes successfully for the all of the databases, including the Asian db.

I have included a reprex set of files for this issue. The are formatted for direct import into the GUI.

Thanks for any insight you can provide.
Keith Inman

Fusion6Cdat_AsianNIST_rev.csv
Suspect_ref.csv
Test_evid.csv
Fusion6Cdat_AfAmNIST_rev.csv

"Marker specific settings" for EFM run on R v4.3.0 leads to a crash (not working).

Error in !is.na(kitdyes) && length(kitdyes) > 1 : 'length = 34' in coercion to 'logical(1)'