Hello all, im working on a cancer gene detection project and im new to bioinformatics so em having a lot of difficulty, i have have to get the normal gene data (nucleotide) from ncbi but the problem is finding the mutated gene sequence related to the normal gene that i have to get from ncbi, i tried to get the data from COSMIC and HGMD but it all failed there was some error when i tried to register from cosmic as it was due to my university which was not in their database same sorta thing happend with hgmd because there was some problem due to the line number where mutation was occuring and when i realte it with the ncbi it was not matching, i registered in hgmd using a public account as i was not able to pay for the professional one, but the thing is i have to do it as it is my final year project for degree completion. please anybody help me it is really important for me.
•
written
by
riazalibaig01 •
0 answers
No answers yet.
Log in to answer this question.
-
RefseqID to gene symbol for multiple species
written by Jacob •
-
How to download TCGA normal sample ?
written by leticia •
-
Where to download Human Gene Mutation Database (HGMD)
written by Shicheng Guo
-
Download Tumor VCF files
written by LimMo •
-
Importing data to excel spreadsheet - gene variant analysis
written by dominik.saburo •
-
Downloading protein sequences for a set of gene IDs from NCBI.
written by Naresh
-
RPKM values in normal cells
written by akantusz •
-
Cancer Datawarehouse
written by dinesh151719 •
-
Discrimination Between Germline And Somatic Mutations In Tumor Without The Availability Of The Norm…
written by Fred Fleche
-
How To Query Hgmd Database In Batch
written by User 4397 •
Can you explain the objective of your project and your approach in detail? Did your advisor tell you that it is feasible and that it makes sense to them?
I ask because your approach seems a lot like the approach I had during my final year of college (there was no exact path to get to where I wanted to go) and I wish someone had asked me for a detailed outline sooner.
sir the objective is A Webserver implementation for cancer gene detection, we need to know if a given gene sequence is mutated or not, we have to compare the normal gene data with the mutated gene data and when the model is completed we need to predict using that ml model we have to predict whether the given sequence given by user is mutated or not which can tell whether the gene is cancerous or not. About the feasibility yes i had talked with advisor basically it was his idea, they had worked on similar project earlier predicting phosphorylation in proteins the method is approximately same at that time i thought it would be easier, but the main issue for me as is the data specially the mutated data which is hard to find im doing this alone and even after spending apprximately month im not successfull in finding usable data
Please talk to your advisor. What you're trying to do has been done in a lot of different ways and in a much more scientifically sound manner. You're trying to recreate TCGA/cBioPortal/COSMIC with none of the scientific underpinnings.
Phosphorylation is a much different topic than mutations. The effect of phosphorylation on proteins is a LOT narrower than the effect of nucleotide changes.
On your current path, you will end up (at best) pointing out somatic mutations (already done) and taking a wild shot at identifying driver genes (done by a better funded, sounder consortium).
Can you tell me your user base for users looking to submit a sequence to find if a gene has somatic mutations? Because users will either have a list of mutations and want to know if any of them are correlated to cancer, or they will have a sequence and will want to know which gene that sequence aligns to.
Also, you speak of a ML model. What will be the training data for that model, and what will you be predicting? What will be the truth set for your input data, which I am guessing is a set of sequences that you say are "cancerous genes"?
i have to not make a platform like TCGA and bioportal etc, it need to be simpler and much focused for a layman user, the user can find his dna sequence from anywhere and when he just enter the sequence on our website it will give answers that wether he/she have cancer or not, i have to use chous 5 step rule that are (1)benchmark data set (2) feature extraction(3)applying ml model(4)cancer prediction (5)webserver implementation
Building something for technical users is A LOT easier than building something for the lay person - you will have to do the analysis AND explain the results to lay people vs just doing a specific analysis for experts.
Your project will be extremely difficult yet produce something that already exists in a better form. I'm telling you this out of experience - I've been in your position and I am currently working on a team that is creating a few ML models on cancer omics datasets. Our effort has taken us 2 years and nearly 10 experts, it's not as simple as you picture it.
Please don't write all words after each other. Make sentences, use punctuation. This is very hard to read.
i will really appreciate if you help me finding the mutated dataset related to the normal data set from ncbi, it is really important for me like as you said i also have no idea where i want to go how i can acheive it , but for completion of my university degree i have to do it and i can't backup either..
You should really choose a different project - you're describing BLAST, VCF annotation and COSMIC to me right now. Your ML model, even if robust, will be <20% effective in any prediction, and that won't be consistent as the role of mutations (in a broader sense) is not well understood across the genome. Mutations can have any effect and they offer a very narrow slit to view the functional world. Phosphoproteomics is a LOT closer to functional entities' changes than mutations are.
I cannot help you beyond this. Even if this were your project and it is too late to start a new one, should you not be further along than just starting out with your dataset search?
we will be using statistical moments calculations, i know that there are already done constoriums and which are in working but i will lose one semster if i have to change the project because it is not an option, all i need mutated data set which i will relate with ncbi data set to see in which line of mutated data set is changed or mutated in relation with the normal data, so when a user enters the gene sequence it will compare it with the already loaded gene sequence that we have and give the result, its right that there are so many types of mutations and nature of mutations also vary but it will be useful if i have even some specific gene mutations for exmaple for BRCA or TP53. i dont have such resources as well as i have no funds to do similar or better than TCGA or cBioportal , but main focus is to find a gene is mutated or not. training and testing data will be the normal nucleotide sequences from ncbi which is available and the mutated data which still i dont have, what should i do if i have to make it possible? even if i have to do for the somatic mutations, im a computer science student and i have not a broader understanding of bio informatics it was also the reason i was not fully aware but i saw bioinformatics interesting also due to the idea which was given to me by supervisor
That is not how ML works. I am really sorry about all the negative feedback, maybe I have tunnel vision on this. I've pointed this thread to other people on the forum, someone should be able to give you some more advice.
Sequence alone cannot give you features for ML. We use expression -omics as well as copy number and small nucleotide changes, and small nucleotide changes are the least informative. Plus, nothing can be predicted at gene level resolution, we need to look at consistent changes between conditions at the pathway level, which cannot be done using mutation data - the effects of mutations cannot be measured in non-trivial cases. Working on mutations is very lucrative especially given how much the word "mutation" is hyped as the cause of bad things, but mutations are a dime a dozen and they are just random changes that in 90% of the cases mean next to nothing.
no problem sir , i have learned many things which i really appreciate, also sir if you find any platform from where i can download mutated dataset please let me know
This project is very ambitious. It would take a team of many highly experienced scientists. I applaud your enthusiasm, but this project doesn't make sense for you to take on.
This doesn't make any sense biologically.