Hi Jason,
Should you spend the $99 to get the BAM file? That depends on what use you're going to make of it.
First, let's understand the different classes of files. Put simply, FASTQ is the set of reads that come out of the DNA sequencer - strings of As, Gs, Cs and Ts (with information about the confidence in those reads) that represent the chopped up bits of DNA that have been read. The BAM format is what you get when you align those reads to a particular reference sequence - normally this is hg38, but it could be realigned to one of the new T2T sequences. The VCF format is what you get when you run through the BAM file to look for differences from that reference sequence - this is what you see in your FTDNA account and what you can download for free; it's very useful for people like me to check the details of variants and whether specific locations are covered.
Now let's consider what you can do with a BAM file that you can't do with a VCF file. Primarily, you can either read it yourself (which most people aren't going to do, because it's fairly technically involved), or you can pay more money to upload it to a third-party comparison site like YFull (€45) or Full Genomes Corp. ($50). You also then have a copy of your DNA results in case of a company-ending calamity at Family Tree DNA.
What do you gain by using these third-party sites? As Wayne points out, you can have your results compared to Y-DNA testers from other companies. This can be really useful for people who have tested outside FTDNA, but FTDNA makes up such a huge proportion of the Y-DNA testing market that you're unlikely to find a close match - however, there's nothing stopping you loading up their phylogenic tree and having a look to see if one exists in your haplogroup already. These sites also give you a "second opinion" on the SNP calls in your test results, including a realignment of your DNA to more up-to-date reference sequences like T2T.
Is that worth it for you, and worth the ~$150 it's going to cost you? It's worth remembering what your goals are. Many people are fixated on a Pokemon-like "got to get them all" attitude towards finding new SNPs. While finding more SNPs can be useful, they are only as useful as the information you can get out of them and the relationships they trace. That means these kind of third-party analyses aren't worth it for most testers, except those who fall into one of a few categories. Some of the most common are:
(1) You have an addiction to cool tools to analyse your DNA, in which case I'd like $100k for a research project. :)
(2) You have checked your haplogroup at YFull and found that you have a haplogroup member outside FTDNA that you really want to be compared to.
(3) You really care about specific STR results in your own test for some reason.
There are all sorts of intangible ramifications from using these third-party tools. The negatives include increasing the exposure of your DNA to malign groups; the positives include making your haplogroup more visible, encouraging competition among DNA-testing companies, etc. All of these are complex in their own right and these kind of BAM downloads / third-party uploads may not be the best way to address any of them. However, historically, when there's been a third-party service that Family Tree DNA have found useful, they've adopted that technology into their own system. We've seen this with Alex Williamson's tree interface (which became the Block Tree), YFull and my TMRCA estimates (which seeded the Discover platform), Dave Vance and Hunter Provyn's SNP Tracker and Phylogeography tools (which, despite the algorithmic limitations, have become co-opted into Globetrekker). Arguably the release of Y-DNA haplogroups from Family Finder tests comes out of tools like MorleyDNA's too. If you think these kind of third-party tools will be useful, then giving them support can eventually make them available to everyone.
?
In summary, the benefit of paying for this download really depends on you having a case for needing it. Speaking personally, I downloaded my BAM file because I have older test and it was free, but I can't say I've made much use of it. I got FGC to analyse my file and report and name my novel variants back when that mattered, but that has less relevance now. There's nothing currently that these third-party sites have been able to offer me that is both of value to me and that I can't get from using a VCF file. Your case may be different, but I'd encourage you to identify an end goal before spending more money.
Cheers,
Iain.
Wayne